Membrane Operations

Membrane Operations Edited by Enrico Drioli and Lidietta Giorno Related Titles Seidel-Morgenstern, A. (ed.) Pereira...

Author: Enrico Drioli | Lidietta Giorno

266 downloads 1338 Views 8MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Membrane Operations

Edited by Enrico Drioli and Lidietta Giorno

Related Titles Seidel-Morgenstern, A. (ed.)

Pereira Nunes, S., Peinemann, K.-V. (eds.)

Membrane Reactors

Membrane Technology

Case Studies to Improve Selectivity and Yields

in the Chemical Industry

2009

ISBN: 978-3-527-31316-7

2006

ISBN: 978-3-527-32039-4

Sammells, A. F., Mundschau, M. V. (eds.) Koltuniewicz, A., Drioli, E.

Membranes in Clean Technologies Theory and Practice 2008

Nonporous Inorganic Membranes for Chemical Processing 2006 ISBN: 978-3-527-31342-6

ISBN: 978-3-527-32007-3

Peinemann, K.-V., Pereira Nunes, S. (eds.)

Membrane Technology

Freeman, B., Yampolskii, Y., Pinnau, I. (eds.)

2007

Materials Science of Membranes for Gas and Vapor Separation

ISBN: 978-3-527-31480-5

2006

Volume 1: Membranes for Life Sciences

ISBN: 978-0-470-85345-0

Peinemann, K.-V., Pereira Nunes, S. (eds.)

Membrane Technology

Ohlrogge, K., Ebert, K. (eds.)

Volume 2: Membranes for Energy Conversion

Membranen

2007

Grundlagen, Verfahren und industrielle Anwendungen

ISBN: 978-3-527-31481-2

2006 ISBN: 978-3-527-30979-5

Li, K.

Ceramic Membranes for Separation and Reaction 2007 ISBN: 978-0-470-01440-0

Membrane Operations Innovative Separations and Transformations

Edited by Enrico Drioli and Lidietta Giorno

The Editors Prof. Enrico Drioli University of Calabria Institute on Membrane Technology Via P. Bucci 17 /C 87030 Rende (CS) Italy

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for

Prof. Lidietta Giorno University of Calabria Institute on Membrane Technology Via P. Bucci 17 /C 87030 Rende (CS) Italy

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliograﬁe; detailed bibliographic data are available on the Internet at http://dnb.d-nb.de. # 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microﬁlm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not speciﬁcally marked as such, are not to be considered unprotected by law. Composition Thomson Digital, Noida, India Printing Betz-Druck GmbH, Darmstadt Bookbinding Litges & Dopf GmbH, Heppenheim Cover Design Formgeber, Eppelheim Printed in the Federal Republic of Germany Printed on acid-free paper ISBN: 978-3-527-32038-7

V

Contents List of Contributors XVII Introduction XXIII

Part One Molecular Separation 1

1.1 1.2 1.3 1.3.1 1.3.2 1.3.2.1 1.3.2.2 1.3.3 1.3.3.1 1.3.3.2 1.3.3.3 1.4

2 2.1 2.2

1

Molecular Modeling, A Tool for the Knowledge-Based Design of Polymer-Based Membrane Materials 3 Dieter Hofmann and Elena Tocci Introduction 3 Basics of Molecular Modeling of Polymer-Based Membrane Materials 5 Selected Applications 7 Hard- and Software 7 Simulation/Prediction of Transport Parameters and Model Validation 8 Prediction of Solubility Parameters 9 Prediction of Diffusion Constants 9 Permeability of Small Molecules and Free-Volume Distribution 12 Examples of Polymers with Low Permeability of Small Molecules (e.g., PO2 50 Barrer) 13 Examples of Polymers with High Permeability of Small Molecules (e.g., 50 Barrer PO2 200 Barrer) 13 Examples of Polymers with Ultrahigh Permeability of Small Molecules (e.g., PO2 1000 Barrer) 14 Summary 16 References 17 Polymeric Membranes for Molecular Separations Heru Susanto and Mathias Ulbricht Introduction 19 Membrane Classiﬁcation 19

19

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

VI

Contents

2.3 2.3.1 2.3.2 2.3.2.1 2.3.2.2 2.3.2.3 2.4 2.4.1 2.4.2 2.4.3 2.4.4 2.5 2.6 2.6.1 2.6.2 2.6.3 2.6.4 2.7

3

3.1 3.2 3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.4 3.4.1 3.4.2 3.4.3 3.4.4 3.5

4 4.1 4.2 4.3 4.4

Membrane Polymer Characteristics 22 Polymer Structure and Properties 22 Membrane Polymer Selection 23 Polymers for Porous Barriers 23 Polymers for Nonporous Barrier 25 Polymers for Charged Barrier 26 Membrane Preparation 26 Track-Etching of Polymer Films 26 Phase Separation of Polymer Solutions 27 Composite Membrane Preparation 30 Mixed-Matrix Membranes 32 Membrane Modiﬁcation 32 Established and Novel Polymer Membranes for Molecular Separations 34 Ultraﬁltration 34 Reverse Osmosis and Nanoﬁltration 36 Pervaporation 37 Separations Using Ion-Exchange Membranes 38 Conclusion and Outlook 40 References 41 Fundamentals of Membrane Solvent Separation and Pervaporation 45 Bart Van der Bruggen Introduction: Separation Needs for Organic Solvents 45 Pervaporation and Nanoﬁltration Principles 46 Membrane Materials and Properties for Solvent Separation 48 Solvent-Stable Polymeric Membrane Materials 48 Ceramic Membrane Materials 49 Solvent Stability 52 Structural Properties for Membranes in NF and PV 52 Flux and Separation Prediction 53 Flux Models in NF 53 Rejection in NF 55 Models for PV: from Solution-Diffusion to Maxwell–Stefan 56 Hybrid Simulations 57 Conclusions 58 References 58 Fundamentals of Membrane Gas Separation 63 Tom M. Murphy, Grant T. Offord, and Don R. Paul Introduction 63 Polymer Structure and Permeation Behavior 64 Membranes from Glassy Polymers: Physical Aging 69 Membranes from Rubbery Polymers: Enhanced CO2 Selectivity

75

Contents

4.5

Summary 79 References 79

5

Fundamentals in Electromembrane Separation Processes 83 Heinrich Strathmann Introduction 83 The Structures and Functions of Ion-Exchange Membranes 84 Ion-Exchange Membrane Materials and Structures 85 Preparation of Ion-Exchange Membranes 85 Preparation Procedure of Heterogeneous Ion-Exchange Membranes 86 Preparation of Homogeneous Ion-Exchange Membranes 86 Special Property Membranes 88 Transport of Ions in Membranes and Solutions 88 Electric Current and Ohms Law in Electrolyte Solutions 89 Mass Transport in Membranes and Solutions 91 The Driving Force and Fluxes in Electromembrane Processes 91 Electrical Current and Fluxes of Ions 91 The Transport Number and the Membrane Permselectivity 92 Membrane Counterion Permselectivity 93 Water Transport in Electrodialysis 94 The Principle of Electromembrane Processes 95 Electrodialysis 95 Electrodialysis System and Process Design 96 Electrodialysis Process Costs 102 Electrodialysis with Bipolar Membranes 107 Electrodialysis with Bipolar Membrane System and Process Design 108 Electrodialysis with Bipolar Membrane Process Costs 110 Continuous Electrodeionization 113 System Components and Process Design Aspects 113 Operational Problems in Practical Application of Electrodeionization 115 Other Electromembrane Separation Processes 115 References 118

5.1 5.2 5.2.1 5.2.2 5.2.2.1 5.2.2.2 5.2.2.3 5.3 5.3.1 5.3.2 5.3.2.1 5.3.2.2 5.3.2.3 5.3.2.4 5.3.2.5 5.4 5.4.1 5.4.1.1 5.4.1.2 5.4.2 5.4.2.1 5.4.2.2 5.4.3 5.4.3.1 5.4.3.2 5.4.4

6 6.1 6.1.1 6.1.2 6.1.3 6.1.4 6.1.5 6.2

Fouling in Membrane Processes 121 Anthony G. Fane, Tzyy H. Chong, and Pierre Le-Clech Introduction 121 Characteristics of Fouling 121 Causes of Fouling 123 Fouling Mechanisms and Theory 125 Critical and Sustainable Flux 125 Fouling and Operating Mode 126 Low-Pressure Processes 126

VII

VIII

Contents

6.2.1 6.2.2 6.2.3 6.2.4 6.2.4.1 6.2.4.2 6.3 6.3.1 6.3.2 6.3.3 6.3.4 6.4

Particulate Fouling 126 Colloidal and Macrosolute Fouling 127 Biofouling and Bioﬁlms 128 Case Studies 128 Water Treatment and Membrane Pretreatment Membrane Bioreactor (MBR) 129 High-Pressure Processes 130 Particulate and Colloidal Fouling 130 Biofouling 132 Scale Formation 133 Cake-Enhanced Osmotic Pressure 135 Conclusions 136 References 136

7

Energy and Environmental Issues and Impacts of Membranes in Industry 139 William J. Koros, Adam Kratochvil, Shu Shu, and Shabbir Husain Introduction 139 Hydrodynamic Sieving (MF and UF) Separations 141 Fractionation of Low Molecular Weight Mixtures (NF, D, RO, GS) 142 Reverse Osmosis – The Prototype Large-Scale Success 144 Energy-Efﬁciency Increases – A Look to the Future 145 Success Stories Built on Existing Membrane Materials and Formation Technology 146 Future Opportunities Relying Upon Developmental Membrane Materials and Formation Technology 149 High-Performance Oleﬁn–Parafﬁn Separation Membranes 149 Coal Gasiﬁcation with CO2 Capture for Sequestration 154 Key Hurdles to Overcome for Broadly Expanding the Membrane-Separation Platform 158 Some Concluding Thoughts 160 References 161

7.1 7.2 7.3 7.4 7.5 7.5.1 7.5.2 7.5.2.1 7.5.2.2 7.6 7.7

8 8.1 8.2 8.2.1 8.2.1.1 8.2.1.2 8.2.1.3 8.2.2 8.3 8.3.1

128

Membrane Gas-Separation: Applications 167 Richard W. Baker Industry Background 167 Current Membrane Gas-Separation Technology 167 Membrane Types and Module Conﬁgurations 168 Hollow Fine Fiber Membranes and Modules 169 Capillary Fiber Membranes and Modules 170 Flat-Sheet Membranes and Spiral-Wound Modules 170 Module Size 170 Applications of Gas-Separation Membranes 171 Nitrogen from Air 171

Contents

8.3.2 8.3.3 8.3.4 8.3.4.1 8.3.4.2 8.3.4.3 8.3.5 8.4 8.4.1 8.4.2 8.4.3 8.4.4 8.5

Air Drying 173 Hydrogen Separation 175 Natural-Gas Treatment 178 Carbon-Dioxide Separation 179 Separation of Heavy Hydrocarbons 182 Nitrogen Separation from High-Nitrogen Gas 182 Vapor/Gas Separations in Petrochemical Operations Future Applications 186 CO2/N2 Separations 186 CO2/H2 Separations 188 Water/Ethanol Separations 189 Separation of Organic Vapor Mixtures 191 Summary/Conclusion 191 References 192

9

CO2 Capture with Membrane Systems 195 Rune Bredesen, Izumi Kumakiri, and Thijs Peters Introduction 195 CO2 and Greenhouse-Gas Problem 195 CO2 Capture Processes and Technologies 196 Membrane Processes in Energy Systems with CO2 Capture 199 Processes Including Oxygen-Separation Membranes 199 Precombustion Decarbonization Processes Including Hydrogen and Carbon Dioxide Membrane Separation 202 Postcombustion Capture Processes with Membrane Separation 205 Properties of Membranes for Hydrogen, Oxygen, and Carbon Dioxide Separation 206 Membranes for Oxygen Separation in Precombustion Decarbonization and Oxy-Fuel Processes 206 Flux and Separation 206 Stability Issues 207 Membranes for Hydrogen Separation in Precombustion Decarbonization 207 Microporous Membranes 208 Dense Metal Membranes 209 Stability Issues 209 Dense Ceramic Membranes 210 Membranes for CO2 Separation in Precombustion Decarbonization 211 CO2 Separation in Postcombustion Capture 211 CO2 Separation Membranes 211 Membrane Contactors for CO2 Capture 212 Challenges in Membrane Operation 212

9.1 9.1.1 9.1.2 9.2 9.2.1 9.2.2 9.2.3 9.3 9.3.1 9.3.1.1 9.3.1.2 9.3.2 9.3.2.1 9.3.2.2 9.3.2.3 9.3.2.4 9.3.3 9.3.4 9.3.4.1 9.3.4.2 9.4

183

IX

X

Contents

9.4.1 9.4.2 9.5

Diffusion Limitation in Gas-Phase and Membrane Support 212 Membrane Module Design and Catalyst Integration 214 Concluding Remarks 216 References 216

10

Seawater and Brackish-Water Desalination with Membrane Operations 221 Raphael Semiat and David Hasson Introduction: The Need for Water 221 Membrane Techniques in Water Treatment 221 Reverse-Osmosis Desalination: Process and Costs 226 Quality of Desalinated Water 228 Environmental Aspects 229 Energy Issues 230 Treatment of Sewage and Polluted Water 232 Membrane Bioreactors 234 Reclaimed Wastewater Product Quality 234 Fouling and Prevention 235 How to Prevent 236 Membrane Cleaning 237 R&D Directions 237 Impending Water Scarcity 237 Better Membranes 237 New Membranes-Based Desalination Processes 238 Summary 240 References 240

10.1 10.2 10.3 10.3.1 10.3.2 10.3.3 10.4 10.4.1 10.4.2 10.5 10.5.1 10.5.2 10.6 10.6.1 10.6.2 10.6.3 10.7

11

11.1 11.1.1 11.2 11.2.1 11.2.2 11.2.3 11.3 11.3.1 11.3.2 11.3.3 11.4 11.4.1 11.4.2 11.5

Developments in Membrane Science for Downstream Processing 245 João G. Crespo Introduction 245 Why Membranes for Downstream Processing? 245 Constraints and Challenges in Downstream Processing 246 External Mass-Transport Limitations 246 Membrane Fouling 247 Membrane Selectivity 249 Concentration and Puriﬁcation of Small Bioactive Molecules 249 Electrodialysis 250 Pervaporation 251 Nanoﬁltration 253 Concentration and Puriﬁcation of Large Bioactive Molecules 255 Ultraﬁltration 256 Membrane Chromatography 260 Future Trends and Challenges 261 References 262

Contents

12 12.1 12.2 12.3 12.4 12.5 12.6

Integrated Membrane Processes 265 Enrico Drioli and Enrica Fontananova Introduction 265 Integrated Membrane Processes for Water Desalination 266 Integrated Membrane Process for Wastewater Treatment 271 Integrated Membrane System for Fruit-Juices Industry 274 Integrated Membrane Processes in Chemical Production 276 Conclusions 281 References 281

Part Two Transformation 13 13.1 13.2 13.3 13.3.1 13.3.2 13.4 13.5 13.6

14 14.1 14.2 14.2.1 14.2.2 14.3 14.3.1 14.3.2 14.3.3 14.4 14.4.1 14.4.2 14.4.3 14.4.4 14.4.4.1 14.4.4.2

285

Fundamental of Chemical Membrane Reactors 287 Giuseppe Barbieri and Francesco Scura Introduction 287 Membranes 289 Membrane Reactors 294 Mass Balance 294 Energy Balance 296 Catalytic Membranes 301 Thermodynamic Equilibrium in Pd-Alloy Membrane Reactor Conclusions 303 References 306

301

Mathematical Modeling of Biochemical Membrane Reactors 309 Endre Nagy Introduction 309 Membrane Bioreactors with Membrane as Bioreactor 310 Enzyme Membrane Reactor 311 Whole-Cell Membrane Bioreactor 312 Membrane Bioreactors with Membrane as Separation Unit 312 Moving-Bed Bioﬁlm Membrane reactor 312 Wastewater Treatment by Whole-Cell Membrane Reactor 313 Membrane Fouling 313 Mathematical Modeling of Membrane Bioreactor 314 Modeling of Enzyme Membrane Layer/Bioﬁlm Reactor 314 Concentration Distribution and Mass-Transfer Rates for Real Systems 318 Prediction of the Convective Velocity through Membrane with Cake and Polarization Layers 321 Convective Flow Proﬁle in a Hollow-Fiber Membrane 323 Without Cake and Polarization Layers 323 With Cake and Polarization Layer 324

XI

XII

Contents

14.4.5 14.5 14.5.1 14.5.2 14.5.3 14.6

15

15.1 15.2 15.2.1 15.2.2 15.2.2.1 15.2.2.2 15.3 15.4 15.4.1 15.4.2 15.4.3 15.4.4 15.4.5 15.4.6 15.5 15.6 15.6.1 15.6.2 15.6.2.1 15.6.2.2 15.6.2.3 15.6.3 15.6.4 15.7 15.7.1 15.7.2 15.8

Mass Transport in the Feed Side of the Hollow-Fiber Membrane Bioreactor 325 Modeling of the MBR with Membrane Separation Unit 327 Moving-Bed-Bioﬁlm Membrane Reactor 327 Submerged or External MBR Process 327 Fouling in Submerged Membrane Module 328 Conclusions and Future Prospects 328 References 332 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds 335 Raffaele Molinari, Angela Caruso, and Leonardo Palmisano Introduction 335 Fundamentals on Heterogeneous Photocatalysis 336 Mechanism 336 Photocatalysts: Properties and New Semiconductor Materials Used for Photocatalytic Processes 336 Titanium Dioxide 338 Modiﬁed Photocatalysts 338 Photocatalytic Parameters 340 Applications of Photocatalysis 341 Total Oxidations 341 Selective Oxidations 343 Reduction Reactions 344 Functionalization 344 Hydrogen Production 345 Combination of Heterogeneous Photocatalysis with Other Operations 346 Advantages and Limits of the Photocatalytic Technologies 346 Membrane Photoreactors 348 Introduction 348 Membrane Photoreactor Conﬁgurations 348 Pressurized Membrane Photoreactors 349 Sucked (Submerged) Membrane Photoreactors 349 Membrane Contactor Photoreactors 350 Parameters Inﬂuencing the Photocatalytic Membrane Reactors (PMRs) Performance 352 Future Perspectives: Solar Energy 353 Case Study: Partial and Total Oxidation Reactions in PMRs 354 Degradation of Pharmaceutical Compounds in a PMR 354 Photocatalytic Production of Phenol from Benzene in a PMR 357 Conclusions 358 References 358

Contents

16 16.1 16.2 16.2.1 16.2.2 16.3 16.3.1 16.3.1.1 16.3.1.2 16.3.2 16.3.3 16.3.3.1 16.3.3.2 16.3.3.3 16.3.4 16.3.4.1 16.3.4.2 16.3.4.3 16.4 16.4.1 16.4.2 16.4.2.1 16.4.2.2

17 17.1 17.2 17.2.1 17.2.2 17.2.3 17.3

18 18.1 18.1.1 18.1.2 18.2 18.2.1 18.2.2

Wastewater Treatment by Membrane Bioreactors 363 TorOve Leiknes Introduction 363 Membranes in Wastewater Treatment 364 Background 364 Membranes Applied to Wastewater Treatment 365 Membrane Bioreactors (MBR) 368 Membrane-Bioreactor Conﬁgurations 368 Membrane Materials and Options 368 Process Conﬁgurations 371 Membrane-Bioreactor Basics 372 Membrane Fouling 374 Understanding Fouling 374 Dealing with Fouling 376 Cleaning Fouled Membranes 378 Deﬁning Operating Conditions and Parameters in MBR Processes 379 Biological Operating Conditions 379 Membrane Filtration Operation 381 Optimizing MBR Operations 383 Prospects and Predictions of the MBR Process 384 Developments and Market Trends 384 An Overview of Commercially Available Systems 386 Flat-Sheet MBR Designs and Options 388 Tubular/Hollow-Fiber MBR Designs and Options 388 References 391 Biochemical Membrane Reactors in Industrial Processes 397 Lidietta Giorno, Rosalinda Mazzei, and Enrico Drioli Introduction 397 Applications at Industrial Level 398 Pharmaceutical Applications 399 Food Applications 402 Immobilization of Biocatalysts on Membranes 405 Conclusion 407 References 407 Biomedical Membrane Extracorporeal Devices 411 Michel Y. Jaffrin and Cécile Legallais General Introduction 411 Use of Membranes in the Medical Field 411 Historical Perspective 411 Hemodialyzers 413 Introduction 413 Physical Principles of Hemodialysis 414

XIII

XIV

Contents

18.2.3 18.2.4 18.2.4.1 18.2.5 18.2.6 18.2.6.1 18.2.6.2 18.3 18.3.1 18.3.2 18.3.3 18.3.3.1 18.3.3.2 18.4 18.4.1 18.4.2 18.4.3 18.4.4 18.4.5 18.4.6

19

19.1 19.2 19.3 19.4 19.5

Dialysis Requirements 415 Mass Transfers in a Hemodialyzer 416 Characterization of Hemodialyzers Performance 416 Hemoﬁltration and Hemodiaﬁltration 417 Various Types of Hemodialyzers 418 Various Types of Membranes 419 Optimization of Hemodialyzer Performance 420 Plasma Separation and Puriﬁcation by Membrane 421 Introduction 421 The Baxter Autopheresis C System for Plasma Collection from Donors 421 Therapeutic Applications of Plasma Separation 422 Plasma Exchange 423 Selective Plasma Puriﬁcation by Cascade Filtration 423 Artiﬁcial Liver 426 Introduction 426 Physical Principles 426 Convection þ Adsorption Systems 428 Diffusion þ Adsorption Systems 428 Future of Artiﬁcial Livers 429 Conclusions 430 References 430 Membranes in Regenerative Medicine and Tissue Engineering 433 Sabrina Morelli, Simona Salerno, Antonella Piscioneri, Maria Rende, Carla Campana, Enrico Drioli, and Loredana De Bartolo Introduction 433 Membranes for Human Liver Reconstruction 434 Human Lymphocyte Membrane Bioreactor 439 Membranes for Neuronal-Tissue Reconstruction 440 Concluding Remarks 443 References 444

Part Three Membrane Contactors 20 20.1 20.2 20.3 20.4 20.5

447

Basics in Membrane Contactors 449 Alessandra Criscuoli Introduction 449 Deﬁnition of Membrane Contactors 449 Mass Transport 452 Applications 455 Concluding Remarks 460 References 460

Contents

21

21.1 21.2 21.3 21.3.1 21.3.2 21.4 21.4.1 21.4.2 21.4.3 21.5 21.5.1 21.5.2 21.5.3 21.5.4 21.6

22 22.1 22.2 22.3 22.4

23

23.1 23.2 23.3 23.3.1 23.4 23.4.1 23.5 23.5.1

Membrane Emulsification: Principles and Applications 463 Lidietta Giorno, Giorgio De Luca, Alberto Figoli, Emma Piacentini, and Enrico Drioli Introduction 463 Membrane Emulsiﬁcation Basic Concepts 465 Experimental Bases of Membrane Emulsiﬁcation 468 Post-Emulsiﬁcation Steps for Microcapsules Production 474 Membrane Emulsiﬁcation Devices 476 Theoretical Bases of Membrane Emulsiﬁcation 479 Torque and Force Balances 480 Surface-Energy Minimization 485 Microﬂuid Dynamics Approaches: The Shape of the Droplets 486 Membrane Emulsiﬁcation Applications 488 Applications in the Food Industry 488 Applications in the Pharmaceutical Industry 489 Applications in the Electronics Industry 490 Other Applications 491 Conclusions 493 References 494 Membrane Contactors in Industrial Applications 499 Soccorso Gaeta Air Dehumidiﬁcation: Results of Demonstration Tests with Refrigerated Storage Cells and with Refrigerated Trucks 505 Refrigerated Storage Cells 507 Refrigerated Trucks 508 Capture of CO2 from Flue Gas 510 References 512 Extractive Separations in Contactors with One and Two Immobilized L/L Interfaces: Applications and Perspectives 513 4 Stefan Schlosser Introduction 513 Contactors with Immobilized L/L Interfaces 516 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS) 517 Case Studies 519 Pertraction through BLME 525 Case Studies 526 Pertraction through SLM 527 Case Studies 529

XV

XVI

Contents

23.6 23.7

Comparison of Extractive Processes in HF Contactors and Pertraction through ELM 529 Outlook 529 References 531 Index

543

XVII

List of Contributors Richard W. Baker Membrane Technology and Research, Inc. 1360 Willow Road Menlo Park, CA 94025 USA Giuseppe Barbieri University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy Loredana De Bartolo University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy Rune Bredesen SINTEF Materials and Chemistry P.O. Box 124 Blindern 0314 Oslo Norway

Carla Campana University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy and University of Calabria Department of Chemical Engineering and Materials Via P. Bucci, cubo 45/A 87030 Rende (CS) Italy Angela Caruso University of Calabria Department of Chemical Engineering and Materials Via P. Bucci, cubo 45/A 87030 Rende (CS) Italy

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

XVIII

List of Contributors

Tzyy H. Chong Nanyang Technological University Singapore Membrane Technology Centre School of Civil and Environmental Engineering Singapore 639798

Anthony G. Fane University of New South Wales UNESCO Centre for Membrane Science & Technology School of Chemical Sciences and Engineering Sydney, NSW 2052 Australia

João G. Crespo Universidade Nova de Lisboa Faculdade de Ciências e Tecnologia Requimte-CQFB Departamento de Química 2829-516 Caparica Portugal

and

Enrico Drioli University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy and University of Calabria Department of Chemical Engineering and Materials Via P. Bucci, cubo 44/A 87030 Rende (CS) Italy Alessandra Criscuoli University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy

Nanyang Technological University Singapore Membrane Technology Centre School of Civil and Environmental Engineering Singapore 639798 Alberto Figoli University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy Enrica Fontananova University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy and University of Calabria Department of Chemical Engineering and Materials Via P. Bucci, cubo 44/A 87030 Rende (CS) Italy

List of Contributors

Soccorso Gaeta GVS S.P.A. Via Roma 50 40069 Zola Predosa (Bo) Italy Lidietta Giorno University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy David Hasson Technion – Israel Institute of Technology Stephen and Nancy Grand Water Research Institute Wolfson Chemical Engineering Department Rabin Desalination Laboratory Technion City Haifa, 32000 Israel Tzyy Haur Nanyang Technological University Singapore Membrane Technology Centre School of Civil and Environmental Engineering Singapore 639798 Dieter Hofmann GKSS Research Center Center for Biomaterial Development of the Institute of Polymer Research Kantstr. 55 14513 Teltow Germany

Shabbir Husain Georgia Institute of Technology School of Chemical & Biomolecular Engineering Atlanta, GA 30332-0100 USA Michel Y. Jaffrin UMR CNRS 6600 Technological University of Compiegne 60200 Compiegne France William J. Koros Georgia Institute of Technology School of Chemical & Biomolecular Engineering Atlanta, GA 30332-0100 USA Adam Kratochvil PRISM Membranes Air Products and Chemicals, Inc. St. Louis, Mo 63146 USA Izumi Kumakiri SINTEF Materials Technology P.O. Box 124 Blindern 0314 Oslo Norway Pierre Le-Clech University of New South Wales UNESCO Centre for Membrane Science & Technology School of Chemical Sciences and Engineering Sydney, NSW 2052 Australia

XIX

XX

List of Contributors

Cécile Legallais UMR CNRS 6600 Technological University of Compiegne 60200 Compiegne France TorOve Leiknes NTNU - Norwegian University of Science and Technology Department of Hydraulic and Environmental Engineering S.P. Andersensvei 5 7491 Trondheim Norway Giorgio De Luca University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy Rosalinda Mazzei University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy and University of Calabria Department of Ecology Via P. Bucci 6/B 87036 Rende (CS) Italy

Raffaele Molinari University of Calabria Department of Chemical Engineering and Materials Via P. Bucci 87030 Rende (CS) Italy Sabrina Morelli University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy T.M. Murphy The University of Texas at Austin Department of Chemical Engineering Austin, TX 78712 USA Endre Nagy University of Pannonia Research Institute of Chemical and Process Engineering P.O. Box 158 8201, Veszprém Hungary Grant T. Offord The University of Texas at Austin Department of Chemical Engineering Austin, TX 78712 USA Leonardo Palmisano University of Palermo Department of Chemical Engineering Processes and Materials Schiavello-Grillone Photocatalysis Group viale delle Scienze 90128 Palermo Italy

List of Contributors

Don R. Paul The University of Texas at Austin Department of Chemical Engineering Austin, TX 78712 USA Thijs Peters SINTEF Materials Technology P.O. Box 124 Blindern 0314 Oslo Norway Emma Piacentini University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy Antonella Piscioneri University of Calabria Institute of Membrane Technology National Research Council of Italy ITM-CNR Via P. Bucci, cubo 17/C 87030 Rende (CS) Italy and University of Calabria Department of Cell Biology via P. Bucci 87030 Rende (CS) Italy

Maria Rende University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy and University of Calabria Department of Chemical Engineering and Materials Via P. Bucci, cubo 45/A 87030 Rende (CS) Italy Simona Salerno University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy 4 Stefan Schlosser Slovak University of Technology Institute of Chemical and Environmental Engineering Radlinského 9 812 37 Bratislava Slovakia Raphael Semiat Technion – Israel Institute of Technology Wolfson Chemical Engineering Department Rabin Desalination Laboratory Stephen and Nancy Grand Water Research Institute Technion City Haifa, 32000 Israel

XXI

XXII

List of Contributors

Francesco Scura University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy

Elena Tocci University of Calabria Institute on Membrane Technology (ITM-CNR) Via P. Bucci, 17/C 87030 Rende (CS) Italy

Shu Shu Georgia Institute of Technology School of Chemical & Biomolecular Engineering Atlanta, GA 30332-0100 USA

Mathias Ulbricht Universität Duisburg-Essen Lehrstuhl für Technische Chemie II 45117 Essen Germany

Heinrich Strathmann University of Stuttgart Institute of Chemical Technology Böblingerstr. 72 70199 Stuttgart Germany Heru Susanto Universität Duisburg-Essen Lehrstuhl für Technische Chemie II 45117 Essen Germany

Bart Van der Bruggen K.U. Leuven, Department of Chemical Engineering Section Applied Physical Chemistry and Environmental Technology W. de Croylaan 46 3001 Heverlee (Leuven) Belgium

XXIII

Introduction Membrane processes are state of the art technologies in various industrial sectors, including gas separation, wastewater treatment, food processing and medical applications. Modelling methodologies are contributing signiﬁcantly to the knowledge-based development of membrane materials and engineering. Micro-ultraﬁltration and reverse osmosis are mature technologies for separations based on molecular exclusion and solution-diffusion mechanisms, respectively. Cleaning and maintenance procedures able to control fouling to an acceptable extent have made these processes commercially suitable. Some of the largest plants for seawater desalination, wastewater treatment and gas separation are already based on membrane engineering. For example, the Ashkelon Desalination Plant for seawater reverse osmosis (SWRO), in Israel, has been fully operational since December 2005 and produces more than 100 million m3 of desalinated water per year. One of the largest submerged membrane bioreactor unit in the world was recently built in Porto Marghera (Italy) to treat tertiary water. The growth in membrane installations for water treatment in the past decade has resulted in a decreased cost of desalination facilities, with the consequence that the cost of the reclaimed water for membrane plants has also been reduced. Membranes are growing signiﬁcantly also in gas separation, for example, the current market size of carbon-dioxide separation from natural gas is more than 70 million Euro/year. Medical applications are among the most important in the membrane market, with hemodialysis, blood oxygenators, plasma separation and fractionation being the traditional areas of applications, while artiﬁcial and bioartiﬁcial organs and regenerative medicine represent emerging areas in the ﬁeld. Nanoﬁltration has achieved a good stage of development, gaining attention in various applications for separations based on both molecular exclusion and charge interaction as well as on the solution-diffusion mechanism. In particular, nanoﬁltration is considered among the most suitable technologies for solvent separation. More recent processes such as membrane reactors, membrane contactors, and membranes in life science are also developing very rapidly. The optimal design of

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

XXIV

Introduction

chemical transformation processes with control of reagent supply and/or product removal through catalytic membranes and membrane reactors is one of the most attractive solutions in process intensiﬁcation. The catalytic action of biocatalysts is extremely efﬁcient, selective and highly stereospeciﬁc when compared to conventional chemical catalysts. Membrane bioreactors are particularly attractive in terms of ecocompatibility, because they do not require additives, are able to operate at moderate temperature and pressure, reduce the formation of by-products, while permitting the production of high valuable coproducts. This may allow challenges in developing new production lines moving towards zero discharge to be faced. The development of catalytic membrane reactors for high-temperature applications became realistic more recently, with the development of high-temperature-resistant membranes. The major market for membrane bioreactors is represented by wastewater treatment with the use of submerged modules conﬁguration. These are considered among the best available technologies by the European Directives on Environment. Membrane bioreactors are also applied in food, red and white biotechnology. In these cases, the external loop conﬁguration is used. Membrane contactors, including membrane crystallizers and membrane emulsiﬁers, are among the most recent membrane operations with growing interest in various industrial sectors. For example, membrane emulsiﬁcation has grown from the 1990s, when it was ﬁrst developed in Japan, to nowadays with applications in food, chemical, pharmaceutical and cosmetic ﬁelds. In Europe, the research at the academic level has achieved a thorough knowledge both from experimental and theoretical points of view. This is fuelling the industrial interest towards the membrane emulsiﬁcation technology, especially for those productions that involve labile bioactive molecules. In general, nowadays the attention towards membrane science and technology is increasing signiﬁcantly. Drivers of this interest include the need for technologies to enable sustainable production, directives and regulations about the use of ecofriendly technologies, consumer demand for high-quality and safe products, public concern about environment, and stakeholder conﬁdence in and acceptance of advanced technologies. Current initiatives recognize that a sustainable solution to the increasing demand of goods and energy is in the rational integration and implementation of new technologies able to achieve concrete beneﬁts for manufacturing and processing, substantially increasing process precision, reducing equipment size, saving energy, reducing costs, and minimizing environmental impact. Membranes and membrane processes are best suited in this context as their basic aspects well satisfy the requirements of process intensiﬁcation for a sustainable industrial production. In fact, they are precise and ﬂexible processing techniques, able to maximize phase contact, integrate conversion and separation processes, with improved efﬁciency and with signiﬁcantly lower energy requirements compared to conventional techniques. This multiauthor book highlights the current state and advances in membranes and membrane operations referring to three major roles of the membrane: mole-

Introduction

cular separation, (bio)chemical transformation and phase contactors. Each topic includes fundamentals and applications of membranes and membrane operations. The largest section is constituted by membranes in molecular separation, which is the most traditional application of membranes. Signiﬁcant advances of membrane science and technologies are expected in transformation processes and membrane contactors for conventional and innovative applications.

XXV

Part One Molecular Separation This Part will be focused on the fundamentals and applications of membranes and membrane operations for separation at the molecular level. Both liquid (including organic solvents) and gaseous streams will be discussed. The book opens with a chapter on molecular modeling to highlight the powerful instruments for designing appropriate membrane materials with predicted properties. This is followed by a chapter on polymeric membranes that discusses the current achievements and challenges on membranes for molecular separation in liquid phase. Subsequent individual chapters discuss membranes in organic solvent separation, gas separation and electrochemical separation. A whole chapter is focused on the fundamentals of fouling molecular separation by membranes are completed by a chapter focused on fouling. and another on energy and environmental issues. The application part of this section illustrates the membrane-assisted molecular separation in (i) gases, with a separate chapter dedicated to the CO2 capture using inorganic membrane; (ii) water desalination; (iii) downstream processing of biological products. A chapter on integrated membrane operations illustrates new strategies in water treatment and chemical production. Membrane separation in the medical ﬁeld has been included in a chapter focused on medical extracorporeal devices, which illustrates the use of membranes for separation of biological ﬂuids and for preparation of bioartiﬁcial organs able to accomplish ex vivo biological transformation (Part headed Transformation). The overall aim of the molecular separation section is to illustrate the current capability of membranes and membrane operations in assisting and governing molecular separations and the future perspectives they offer for a more sustainable industrial growth through innovative process design. Their implementation will lead to concrete beneﬁts in manufacturing and processing, substantially shrinking equipment size, boosting plant efﬁciency, saving energy, reducing capital costs, minimizing environmental impact, and using remote control and automation.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

Membrane operations have the potential to replace conventional energy-intensive separation techniques, such as distillation and evaporation, to accomplish the selective and efﬁcient transport of speciﬁc components, to improve the performance of reactive processes and, ultimately, to provide reliable options for a sustainable industrial growth. This is in line with the strategy of process intensiﬁcation and it is expected to bring substantial improvements in chemical and many other manufacturing and processing industries. Many membrane operations are based on similar materials and structures, while differing in the method by which they carry out the separation process. Step forward innovations can be promoted by appropriate integration of traditional membrane operations (reverse osmosis, micro-, ultra- and nanoﬁltration, electrodialysis, pervaporation, etc.) among them and with innovative membrane operations. In fact, while being already widely used in many different applications, they can be combined with new membrane systems such as catalytic membrane reactors and membrane contactors. Nowadays, redesign of industrial production cycles by combining various membrane operations suitable for separation, conversion and concentration units is an attractive opportunity because of the synergic effects that highly integrated membrane processes can promote.

j3

1 Molecular Modeling, A Tool for the Knowledge-Based Design of Polymer-Based Membrane Materials Dieter Hofmann and Elena Tocci

1.1 Introduction

Most important macroscopic transport properties (i.e., permeabilities, solubilities, constants of diffusion) of polymer-based membranes have their foundation in microscopic features (e.g., free-volume distribution, segmental dynamics, distribution of polar groups, etc.) which are not sufﬁciently accessible to experimental characterization. Here, the simulation of reasonably equilibrated and validated atomistic models provides great opportunities to gain a deeper insight into these microscopic features that in turn will help to develop more knowledge-based approaches in membrane development. The mentioned transport properties for small and medium-sized molecules in polymers are decisive in many technologically important processes, for example, in biotechnology and biomedicine, in pharmacological and chemical industries but also in integrated environmental protection. The respective penetrants can be anything from rather small hydrogen or oxygen molecules to chemicals like benzene up to relatively large drug molecules. Membrane processes for the separation of gaseous and liquid mixtures are important examples. In these cases there are already large numbers of applicable materials and processes. Further improvements (mostly concerning better selectivities at acceptably high permeabilities), often needing real jumps in performance, are, however, still needed in many cases. This applies, although in the opposite sense, also to barrier materials where permeations at least of certain types of molecules will be extremely small. Other areas concern biomaterials or material systems for the controlled release of drugs. More speciﬁc examples for the need to develop new materials with tailored transport properties are: .

The separation of methane from higher hydrocarbons in natural gas for safer and more economical transport through pipelines, or for better exploitation;

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

4

.

.

. . . . . . . . .

The design of packaging materials for conservation of fresh fruits and vegetables, which means good speciﬁc permeation and selectivity properties in order to maintain a modiﬁed/controlled atmosphere; The control of migration of additives, monomers or oligomers, from packaging materials, for example, into food (important for the enforcement of a high level of food quality and safety) or other consumer products; The resistance of resins used in composites for aircraft construction to ageing caused by water absorption; Small but continuous fuel loss by permeation through polymeric parts of the fuel system; Separation of CO2 from ﬂue gases, and separation of NOx from vehicle emissions; Efﬁcient and inexpensive proton-conducting membranes for fuel cells; components in polymer electronics (such as for light-emitting diodes or display components) with extremely low permeabilities for oxygen and water; Optimum controlled drug release systems, for example, for medical applications, cosmetics or agriculture; Transport problems in artiﬁcial or bio-hybrid organs; Optimum biocompatibility of polymers in contact with cells and blood; Optimum chemical degradation behavior (often to a large extent a water-permeation problem) for surgical sutures, scaffold materials for tissue engineering, degradable screws in orthopaedic surgery and so on.

In the near future, the use of multifunctional polymer-based materials with separation/selective transport capabilities is also to be expected in the design of production systems with integrated environmental protection or in the combination of chemical reactions and separation by attaching a catalytic functionality to the respective material [1]. Thus, those multifunctional materials should contribute materially to the development of clean energy and/or energy saving and therefore sustainable production technologies. In connection with these perspectives, there is considerable interest in new/modiﬁed polymer-based materials with tailored transport/catalytic properties. Also, many sensor applications are based on controlled permeation. Amorphous polymers or respective composites with inorganic components are an important class of materials to solve many of the above-mentioned problems. However, the design of these multifunctional materials, based on experimentation and correlative thinking alone is unreliable, time consuming, expensive and often not successful. Systematic multiscale computer-aided molecular design (CAMD) offers a very attractive alternative, insofar as these techniques allow for the very elaborate investigation of complex material behavior with regard to the links between structure, dynamics and relevant properties of the discussed multifunctional polymer-based materials on the length and time scales (from Angstroms to micrometers and from picoseconds to milliseconds, respectively) which are most important for the penetrant transport and other relevant processes (e.g., selective transport, separation, catalysis, biodegradation, sensor applications) of interest. In the present chapter, molecular modeling tools (i.e., quantum chemistry (QM), atomistic- and mesoscale modeling) will be in the focus of interest. Consequently, the microscopic properties to

1.2 Basics of Molecular Modeling of Polymer-Based Membrane Materials

be related with macroscopically determined transport parameters are, for example, chain stiffness parameters, free volume and its distribution, mobility measures for chain segments, energy densities describing interactions of chain segments with penetrants, microscopic effects of swelling and so on. Over the last 15 years particularly atomistic molecular modeling methods have found widespread application in the investigation of small-molecule permeation [2–15].

1.2 Basics of Molecular Modeling of Polymer-Based Membrane Materials

The permeation of small molecules in amorphous polymers is typically following the solution diffusion model, that is, the permeability Pi of a feed component i can be envisioned as the product of the respective solubility Si and constant of diffusion Di. Both parameters can be obtained experimentally and in principle also by atomistic simulations. The molecular modeling of these polymers typically starts with the construction of normally rectangular packing models. There, the related chain segments of the respective polymer will be arranged in realistic, that is, statistically possible, way. To do this, ﬁrst the involved atoms are considered to be spheres of the respective atomic radius Ri (as obtainable from QM) and atomic weight mi. The bonded interactions between atoms resulting in bonds, bond angles and conformation angles are then described by mechanic springs or torsion rods with spring constants related to, for example, experimentally known bond strengths. So-called nonbond interactions between atoms that either belong to different molecules or that in one and the same molecule are further apart from each other than about three bonds are considered via, for example, Lennard-Jones (to describe van der Waals interactions) and Coulomb potentials (to describe electrostatic interactions). The sum of all interatomic interactions written as the potential energy of a packing model is then called a forceﬁeld. Forceﬁelds form the core of all atomistic molecular modeling programs. Equation 1.1 shows the principal structure of a typical forceﬁeld for a system of N atoms with the Cartesian atomic position vectors ~ ri . X X V~ r1 ;~ r2 ; . . . ;~ rN ¼ Kb ðll0 Þ2 þ KQ ðQQ0 Þ2 Covalent bonds

X

þ

Bond angles

Kj ½1 þ cosðnjdÞ

Dihedral angles

X

þ

nonbonded atom pairs i;j

with the following parameters:

"

! ! # aij bij qi qj 6 þ e0 er rij rij12 rij

ð1:1Þ

j5

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

6

l ¼ actual length of a bond l0 ¼ length of a bond in equilibrium Kb ¼ force constant for a bond length deformation Q ¼ actual value for a bond angle Q0 ¼ value for a bond angle in equilibrium KQ ¼ force constant for a bond-angle deformation j ¼ actual value for a conformation angle n ¼ periodicity parameter in a conformation potential d ¼ constant to ﬁx trans-state in a conformation potential Kj ¼ force constant for a conformation potential Rij ¼ distance between atoms i and j with (j i) > 3 aij ¼ constant describing repulsive interactions in the Lennard-Jones Potential bij ¼ constant describing attractive interactions in the Lennard-Jones Potential qi ¼ partial charge of the ith atom e0 ¼ vacuum permittivity er ¼ dielectric constant. The parameters l0, Kb, Q0, KQ, Kj, n, d, aij, bij, qi, qj and er belong to the ﬁt parameters, which can be determined by ﬁtting of Equation 1.1 to a sufﬁcient set of data calculated by QM and/or determined experimentally (e.g., X-ray scattering, IR spectroscopy, heats of formation). From a numeric point of view the pair interaction terms (van der Waals and Coulomb) are most demanding. In this connection the typical size of polymer packing models is limited to typically 3000–10 000 atoms (leading to lateral sizes of bulk models of a few nm), although in other connections now also models with up to 100 000 atoms have been used. Forceﬁelds may be utilized in two directions: Model systems can be, on the one hand, subjected to a static structure optimization. There, the fact is considered that the potential energy of a relaxed atomistic system (cf. Equation 1.1) should show a minimum value. Static optimization then means that by suited numeric procedures the geometry of the simulated system is changed as long as the potential energy reaches the next minimum value [16]. In the context of amorphous packing models, the main application for this kind of procedure is the reduction of unrealistic local tensions in a model as a prerequisite for later molecular dynamic (MD) simulations. It is, on the other hand, possible to use the potential energy of a model system as described by Equation 1.1 to calculate the forces ~ Fi acting on each atom of the model via the gradient operation: qVð~ r1 ;~ r2 ; . . . ;~ rN Þ ~ Fi ¼ q~ ri

ð1:2Þ

Then, Newtons equations of motion can be solved for every atom of the investigated system: ri ðtÞ d~ ~ Fi ¼ mi dt2 2

ð1:3Þ

1.3 Selected Applications

The necessary starting positions ~ ri ð0Þof the atoms are in the given case usually obtained from methods of chain-packing procedures (see below). The starting velocities ~ vi ð0Þ of all atoms are assigned via a suited application of the well-known relation between the average kinetic Ekin energy of a polyatomic system and its temperature T:

Ekin ¼

N X 1 i¼1

2

mi~ vi 2 ¼

3N6 kb T 2

ð1:4Þ

kB istheBoltzmannconstant.(3N 6)isthenumberofdegreesoffreedomofanN-atom model considering the fact that in the given case the center of mass of the whole model with its 6 translation and rotation degrees of freedom does not move during the MD simulation. Using Equations 1.2–1.4 it is then possible to follow, for example, the motions of the atoms of a polymer matrix and the diffusive movement of imbedded small penetrant molecules at a given temperature over a certain interval of time. Equation 1.3 represents a system of usually several thousand coupled differential equations of second order. It can be solved only numerically in small time steps Dt via ﬁnite-difference methods [16]. There always the situation at t þ Dt is calculated from the situation at t. Considering the very fast oscillations of covalent bonds, Dt must not be longer than about 1 fs to avoid numerical breakdown connected with problems with energy conservation. This condition imposes a limit of the typical maximum simulation time that for the above-mentioned system sizes is of the order of several ns. The limited possible size of atomistic polymer packing models (cf. above) together with this simulation time limitation also set certain limits for the structures and processes that can be reasonably simulated. Furthermore, the limited model size demands the application of periodic boundary conditions to avoid extreme surface effects. The already mentioned limited lateral dimensions of packing models of just several nm makes it impossible to simulate complete membranes or other polymerbased samples. Therefore, on the one hand, bulk models are considered that are typically cubic volume elements of a few nanometers side length that represent a part cut out of the interior of a polymer membrane (cf. Figure 1.1). On the other hand interface models are utilized, for example, for the interface between a liquid feed mixture and a membrane surface or between a membrane surface and an inorganic ﬁller (cf. Figure 1.2).

1.3 Selected Applications 1.3.1 Hard- and Software

The InsightII/MaterialsStudio/Discover software of Accelrys [18, 19] was utilized for the amorphous packing model construction, equilibration and the atomistic

j7

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

8

Figure 1.1 Atomic representation of a typical 3-dimensional packing model (thickness about 3 Å) starting with a single Hyflon AD60X polymer chain. Atom colors: gray ¼ carbon, red ¼ oxygen, light blue ¼ fluorine [15].

simulations. In most of the following examples the COMPASS forceﬁeld was applied [20, 21]. For data evaluation also self-programmed software (mostly in BTCL, Fortran, C) was applied. Data production runs were performed on a 74 processor Opteron Linux Cluster, a SGI Origin 2100 and on SGI Onyx workstation. 1.3.2 Simulation/Prediction of Transport Parameters and Model Validation

The quality of atomistic packing models is typically validated via comparisons between measured and simulated properties like wide-angle X-ray scattering (WAXS)

Figure 1.2 Atomic representation of a surface model of Pebax/30%KET with water [17].

1.3 Selected Applications

curves, densities, transport parameters for small and medium sized penetrants. In the latter case both validating (if a polymer is already existing and experimentally characterized) and predictive (if a polymer has not been synthesized yet or if no transport parameters are available experimentally) applications are possible. 1.3.2.1 Prediction of Solubility Parameters Here, hitherto in most cases the transition-state method of Gusev and Suter [22, 23] was utilized to ﬁrst determine calculated solubility values Scalc values. There, a ﬁne 3D-grid with a grid spacing of about 0.03 nm is layered over a completely reﬁned detailed-atomistic amorphous polymer bulk packing model (cf. Figure 1.1). Then a small virtual test molecule of the intended kind (e.g., O2) in a united atom representation is inserted in the polymer matrix at each lattice point of the grid. The resulting nonbonded interaction energy Eins between the inserted molecule and the whole polymer matrix is calculated for each position of the respective inserted molecule. Only the van der Waals interactions are considered, that is, the method would not work for highly polar penetrants like water. Furthermore, since the polymer matrix can not locally relax to accommodate larger inserted penetrants it only works for small molecules (typically just up to O2, N2, etc.). From the insertion energy data via Equation 1.5 the chemical excess potential mex for inﬁnite dilution can be calculated and converted in the respective solubility using Equation 1.6.

mex ¼ RT ln < expðEins =kTÞ > Scalc ¼

m T0 exp ex p0 T kT

ð1:5Þ ð1:6Þ

with R being the universal gas constant and T0 and p0 being temperature and pressure under standard conditions (T0 ¼ 273.15 K; p0 ¼ 1013 105 Pa). Table 1.1 contains typical solubility prediction data for an ultrahigh free-volume polymer (PTMSP) and a polymer with more conventional transport properties (PTMSS). As already mentioned the Gusev–Suter method normally only works for small penetrant molecules like oxygen or nitrogen. For a long time no really generally applicable alternative method was available to overcome the problem, but a few years ago Boulougouris, Economou Theodorou et al. [27, 28] suggested a new inverse Widom method based on the particle-deletion algorithm DPD to overcome this problem in principle. The related computer code was, however, only applicable to special, relatively simple model systems. Based on DPD also a generalized version of this algorithm was presented in the literature [29] permitting the calculation of solubility coefﬁcients for molecules as large as, for example, benzene in polymers for which reasonable forceﬁeld parameters exist. Table 1.2 contains solubility data for a number of penetrants of different size in PDMS obtained in this way. 1.3.2.2 Prediction of Diffusion Constants The following description again follows the already quoted papers of Gusev and Suter. Using the Eins values mentioned in the foregoing section, the whole packing model in

j9

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

10

Table 1.1 Results of application of the Gusev–Suter method to the

solubility of N2 in PTMSP and PTMSS.

Polymer Structure formula

Average simulated N2 solubility Average measured N2 solubility coefficient coefficient Scalc [cm3(STP)/(cm3 atm)] Sexp [cm3(STP)/(cm3 atm)]

PTMSP

1.16 [24]

1.02 [25]

PTMSS

0.19 [24]

0.18 [26]

question is separated into regions of free volume (low interaction energy) and regions of densely packed polymer (high interaction energy; cf. Figure 1.3). The borders between the energetically attractive regions Eins(x, y, z) around the resulting local insertion energy minima are given as crest surfaces of locally maximum insertion energy. In the two-dimensional analogy of a cratered landscape a minimum energy region would be represented by a crater, while the crest surface of locally maximum insertion energy would be reduced to the crest line separating one crater from the adjacent ones. From this identiﬁcation of energetically separated sites where a penetrant would typically sit (approximately the centers of holes) and jump probabilities between adjacent sites (which can be calculated by proper integration over the mentioned crest lines and craters of the insertion energy function Eins(x, y, z) an efﬁcient Monte Carlo simulation method for the jump-like diffusion of small

1.3 Selected Applications Table 1.2 Results of application of a generalized DPD method to different penetrants in PDMS.

Solute

Scalc [cm3(STP) cm3 bar1]

Sexp [cm3(STP) cm3 bar1]

Oxygen Nitrogen Acetone Benzene

0.32a 0.13a 69a 495a

0.224b 0.127b 33–66c 275–624d

a

[29, 30]. [31]. c [32]. d [33]. b

molecules in a polymer matrix can be developed (cf. Figure 1.4). With this algorithm the simulation range can almost extend in the ms range. That is, in most cases the normal diffusive regime can be reached and the respective constant of diffusion Di can be obtained via the Einstein equation from the slope of the mean squared displacement si(t): ~ ri ðtÞ~ ri ð0Þj2 i si ðtÞ ¼ h~

! Di ðtÞ ¼

hj~ ri ðtÞ~ ri ð0Þj2 i 6t

ð1:7Þ

ð1:8Þ

Here,~ ri ðtÞis the position vector of penetrant i and <> is the average over all possible time origins t ¼ 0 and all simulated trajectories of a penetrant of a given kind. Again, as with the solubilities the Gusev–Suter method can only handle small penetrants in this way, because the respective polymer matrix cannot conformationally adjust to larger penetrants. Table 1.3 contains a comparison between experimental and

Figure 1.3 Free volume for a perfluorinated polymer in red indicating into the densely packed polymer.

j11

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

12

Figure 1.4 Jump-like diffusion of oxygen molecules in a perfluorinated polymer matrix.

calculated values, Dexp and Dcalc, respectively for a number of gases in PTMSP. Here, for methane and carbon dioxide it has to be considered that these molecules are normally already too large to lead to reasonable results with the Gusev–Suter method. In comparing simulated and experimentally measured transport parameters one has to be aware that experimental data in the literature depending, for example, on sample preparation conditions and the chosen measurement methodology can show a considerable scatter, often reaching a factor of two or even more. It is, for example well-known that polyimides often contain residual solvent ﬁlling a part of the free volume and thus leading to systematically lower S and D values from experiments than from simulations [34]. 1.3.3 Permeability of Small Molecules and Free-Volume Distribution

The distribution of free volume in amorphous polymers is of paramount importance for the respective materials transport behavior towards small and medium-sized penetrants. Table 1.3 Results of application of the Gusev–Suter method for

the diffusion constants of different penetrants in PTMSP. Solute

Dcalc [105 cm2/s]

Dexp [105 cm2/s]

Nitrogen Oxygen Methane Carbon dioxide

7.7a 7.5a 8.2a 9.2a

3.50b 4.66b 2.64b 8.02b

a

[24]. [25].

b

1.3 Selected Applications

While in rubbery polymers differences in the segmental mobility can be more important than differences in the free-volume distribution for glassy polymers often certain basic correlations can be found between the permeability of small molecules and free-volume distribution. Other important factors are the molecular mobility of chain segments and the local chemical composition. Experimentally, the free-volume distribution can be best characterized with positron annihilation lifetime spectroscopy (PALS). There, in organic glasses ortho-positronium (o-Ps) which has a lifetime of 142 ns in vacuo shows a strong tendency to localize in heterogeneous regions of low electron density (holes). In polymeric materials the vacuum lifetime is cut short via the pick-off mechanism, where o-Ps prematurely annihilates with one of the surrounding bound electrons. This lifetime can (under certain assumptions) be converted in an average hole radius [35, 36], while the intensity of the lifetime signal may permit conclusions about the overall contents of free volume. There are, however, a number of shortcomings with common PALS methodology. Often, the holes forming the free volume are assumed to be just spheres and the shape of calculated hole radius distribution peaks is set to Gaussian. Furthermore, positrons in their limited lifetime seem not to be capable of probing large holes of complex topology (cf. in particular PTMSP and other ultrahigh free-volume polymers) [24, 37]. Finally the size of the positronium molecule does only permit probing of the accessible free volume for molecules about the size of hydrogen. Atomistic molecular modeling utilizing bulk models on the other hand can provide additional even more detailed information about free-volume distributions in amorphous polymers. In this way, glassy polymers, where individual differences in chain segment mobility do not have an as distinct inﬂuence on transport properties than in rubbery polymers, can be roughly grouped into three classes regarding their small molecule permeability, as will be outlined in the following for the example of oxygen. 1.3.3.1 Examples of Polymers with Low Permeability of Small Molecules (e.g., PO2 50 Barrer) Figure 1.5(a) shows as a typical example a computer-tomography-like atomic monolayer representation of a bulk model for diisopropyldimethyl PEEK WC (DIDMPEEK). In this case the oxygen-accessible free volume is obviously organized in relatively small isolated holes and the respective size distribution (cf. Figure 1.5(b)) is monomodal and extending only to hole radii of about 5 Å. 1.3.3.2 Examples of Polymers with High Permeability of Small Molecules (e.g., 50 Barrer PO2 200 Barrer) Similarly to Figure 1.5(a), Figure 1.6(a) displays an atomic monolayer representation for a so-called high-performance polymer (here PPrSiDPA with a PO2 of 230 Barrer [38]). Already in this view larger holes are visible than for the case of low-performance polymers (cf. Figure 1.5(a)) and the hole-size distribution (Figure 1.6(b)) reveals a much wider range of radii (here extending to 10 Å and being bi-modal). This situation is quite typical for polymers with high gas transport capacity. A more systematic study

j13

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

14

Figure 1.5 (a) Atomic monolayer representation (thickness about 3 Å) of a typical packing model and structure formula for DIDMPEEK. (b) Hole-size distribution for the packing model shown in Figure 1.5(a).

on polyimides [34] did, for example, reveal that the major difference between lowperformance and high-performance polyimides with about the same overall contents of free volume lies in the distribution of the (e.g., oxygen) accessible free volume. Low-performance polyimides show just a monomodal distribution extending up to about 5 Å, while high-performance polyimides behave more or less similar to the example illustrated in Figure 1.6. 1.3.3.3 Examples of Polymers with Ultrahigh Permeability of Small Molecules (e.g., PO2 1000 Barrer) Figure 1.7 then shows respective data for an ultrahigh free-volume and performance polymer, Teﬂon AF2400 of DuPont (PO2 ¼ 1140 Barrer; [39]). One can recognize that

1.3 Selected Applications

Figure 1.6 (a) Atomic monolayer representation (thickness about 3 Å) of a typical packing model and structure formula for PPrSiDPA. (b) Hole-size distribution for the packing model shown in Figure 1.6(a).

in this case there is conventional free volume organized in isolated holes in the radius range below 10 Å existing in parallel with a partly continuous phase of much larger holes that in this case are visible as a peak between 15 and 20 Å. The effect is even more pronounced for PTMSP, the polymer of this kind with the highest oxygen permeability so far measured (about 9000 Barrer; [38]). There, the continuity for the large-hole phase is more clearly visible already in atomic monolayer representations of respective packing models [37] and the ratio between the area under the conventional free-volume peak and the continuous hole phase peak in the holesize distribution is even smaller than for Teﬂon AF2400. The fact that for the mentioned ultrahigh free-volume polymers the continuous hole-phase peak appears at rather limited values is related with the limited size of the

j15

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

16

Figure 1.7 (a) Atomic monolayer representation of a typical packing model and structure formula for Teflon AF2400. (b) Holesize distribution for the packing model shown in Figure 1.7(a).

investigated models (45–50 Å) while the thickness of real polymer membranes can extend into the micrometer range. 1.4 Summary

Atomistic molecular modeling techniques have proven to be a very useful tool for the investigation of the structure and dynamics of dense amorphous membrane polymers and of transport processes in these materials. By utilizing these methods, information can be obtained that is not accessible by experimental means.

References

Acknowledgments

The work was in part supported by the European projects Growth Program, PERMOD – Molecular modeling for the competitive molecular design of polymer materials with controlled permeability properties. Contract #G5RD-CT-2000-200, the 6 Framework Programme project MULTIMATDESIGN- Computer aided molecular design of multifunctional materials with controlled permeability properties, Contract no.: 013644, INTAS - RFBR 97–1525 grant.

References 1 Drioli, E., Natoli, M., Koter, I. and Trotta, F. (1995) Biotechnology and Bioengineering, 46, 415. 2 Boyd, R.H. and Krishna Pant, P.V. (1991) Macromolecules, 24, 6325. 3 M€ uller-Plathe, F. (1991) Journal of Chemical Physics, 94, 3192. 4 Sok, R.M., Berendsen, H.J.B. and van Gunsteren, W.F. (1992) Journal of Chemical Physics, 96, 4699. 5 Krishna Pant, P.V. and Boyd, R.H. (1993) Macromolecules, 26, 679. 6 Gusev, A.A., M€ uller-Plathe, F., van Gunsteren, W.F. and Suter, U.W. (1994) Advances in Polymer Science, 16, 207. 7 Tamai, Y., Tanaka, H. and Nakanishi, K. (1994) Macromolecules, 27, 4498. 8 Chassapis, C.S., Petrou, J.K., Petropoulos, J.H. and Theodorou, D.N. (1996) Macromolecules, 29, 3615. 9 Fried, J.R., Sadad-Akhavi, M. and Mark, J.E. (1998) Journal of Membrane Science, 149, 115. 10 van der Vegt, N.F.A., Briels, W.J., Wessling, M. and Strathmann, H. (1999) Journal of Chemical Physics, 110, 11061. 11 Hofmann, D., Fritz, L., Ulbrich, J., Schepers, C. and B€ohning, M. (2000) Macromolecular Theory and Simulations, 9, 293. 12 Tocci, E., Hofmann, D., Paul, D., Russo, N. and Drioli, E. (2001) Polymer, 42, 521. 13 Neyertz, S. and Brown, D. (2001) Journal of Chemical Physics, 115, 708; Neyertz, S. Brown, D. Douanne, A. Bas, C. and Alberola, N.D. (2002) The Journal of Physical Chemistry. B, 106, 4617.

14 van der Vegt, N.F.A. (2002) Journal of Membrane Science, 205, 125; van Gunsteren, W.F. and Berendsen, H.J.C. (1990) Angewandte Chemie, 29, 992. 15 Macchione, M., Jansen, J.C., De Luca, G., Tocci, E., Longeri, M. and Drioli, E. (2007) Polymer, 48, 2619–2635. 16 Haile, J.M. (1992) Molecular Dynamics Simulation Elementary Methods, Wiley Interscience, New York. 17 Tocci, E., Gugliuzza, A., De Lorenzo, L. and Drioli, E. in preparation. 18 (1999) Polymer User Guide, Amorphous Cell Section, Version 400p þ , Molecular Simulations Inc., San Diego, CA. 19 (1999) Discover User Guide, Polymer User Guide, Amorphous Cell Section, Version 4.0.0p þ , Molecular Simulations Inc., San Diego, CA. 20 Sun, H. and Rigby, D. (1997) Spectrochim Acta, 53A, 1301. 21 Rigby, D. Sun, H. and Eichinger, B.E. (1997) Polymer International, 44, 311. 22 Gusev, A.A. Arizzi, S. and Suter, U.W. (1993) Journal of Chemical Physics, 99, 2221. 23 Gusev, A.A. and Suter, U.W. (1993) Journal of Chemical Physics, 99, 2228. 24 Hofmann, D., Heuchel, M., Yampolskii, Yu., Khotimskii, V. and Shantarovich, V. (2002) Macromolecules, 35, 2129–2140. 25 Ichiraku, Y., Stern, S.A. and Nakagawa, T. (1987) Journal of Membrane Science, 34, 5–18; Masuda, T., Iguchi, Yu., Tang, B.-Z. and Higashimura, T. (1988) Polymer, 29, 2041–2049; Bondar, V.I. (1995) Ph.D.

j17

j 1 Molecular Modeling, A Tool for the Knowledge-Based Design Design of Polymer-Based

18

26

27

28

29

30

31

32

Thesis, Institute of Petrochemical Synthesis, Moscow. Khotimskii, V.S., Filippova, V.G., Bryantseva, I.S., Bondar, V.I., Shantarovich, V.P. and Yampolskii, Y.P. (2000) Journal of Applied Polymer Science, 78, 1612–1620. Boulougouris, G.C., Economou, I.G. and Theodorou, D.N. (2001) Journal of Chemical Physics, 115, 8231–8237. Boulougouris, G.C., Voutsas, E.C., Economou, I.G., Theodorou, D.N. and Tassios, D.P. (2001) The Journal of Physical Chemistry B, 105, 7792–7798. Siegert, M.R., Heuchel, M. and Hofmann, D. (2007) Journal of Computational Chemistry, 28, 877–889. Simulated data from, Siegert, M.R. (2006) PhD Thesis, Berechnung von L€ oslichkeitskoefﬁzienten in PolymerMaterialien, FU-Berlin. Kamiya, Y., Naito, Y., Hirose, T. and Mizoguchi, K. (1990) Journal of Polymer Science, Part B, Polymer Physics, 28 (8), 1297–1308. Singh, A., Freeman, B.D. and Pinnau, I. (1998) Journal of Polymer Science, Part B, Polymer Physics, 36 (2), 289–301; Frahn, J. (GKSS), personal communication.

33 Summers, W.R., Tewari, Y.B. and Schreiber, H.P. (1972) Macromolecules, 5 (1), 12–16; Lichtenthaler, R.N., Liu, D.D. and Prausnitz, J.M. (1974) Berichte der Bunsen-Gesellschaft — Physical Chemistry Chemical Physics, 78 (5), 470–477; Grate, J.W., Kaganove, S.N. and Bhethanabotla, V.R. (1997) Faraday Discussions, 107, 259–283; Frahn, J. (GKSS), personal communication. 34 Heuchel, M., Hofmann, D. and Pullumbi, P. (2004) Macromolecules, 37, 201. 35 Gregory, R.B. and Yongkang, Zhu (1991) Positron and Positron Chemistry (ed. Y.C. Jean), World Scientiﬁc, Singapore, p. 136. 36 Provensher, S.W. (1982) Computer Physics Communications, 27, 229. 37 Hofmann, D., Entrialgo-Castano, M., Lerbret, A., Heuchel, M. and Yampolskii, Yu. (2003) Macromolecules, 36, 8528–8538. 38 Yampolskii, Yu.P., Korikov, A.P., Shantarovich, V.P., Nagai, K., Freeman, B.D., Masuda, T., Teraguchi, M. and Kwak, G. (2001) Macromolecules, 34, 1788–1796. 39 Alentiev, A.Yu., Yampolskii, Yu.P., Shantarovich, V.P., Nemser, S.M. and Plate, N.A. (1997) Journal of Membrane Science, 126, 123–132.

j19

2 Polymeric Membranes for Molecular Separations Heru Susanto and Mathias Ulbricht

2.1 Introduction

In this chapter we describe the state-of-the-art and the challenges in preparation and manufacturing of polymeric membranes for molecular separations in liquid phase. The processes include separation of aqueous solutions, that is, pressure-driven desalination using reverse osmosis and nanoﬁltration, fractionations of small and larger molecules using ultraﬁltration and removal of organic substances by pervaporation (e.g., for shifting equilibria for (bio)chemical reactions). Separations in nonaqueous organic systems such as pervaporation and nanoﬁltration will also be covered. The preparation of charged membranes for electromembrane processes is another important application area for special polymers. Surface modiﬁcation of membranes has become an important tool to reduce fouling or increase biocompatibility, but it can also be used to change membrane selectivity by combining separation mechanisms (e.g., based on size and charge).

2.2 Membrane Classification

Synthetic membranes for molecular liquid separation can be classiﬁed according to their selective barrier, their structure and morphology and the membrane material. The selective barrier – porous, nonporous, charged or with special chemical afﬁnity – dictates the mechanism of permeation and separation. In combination with the applied driving force for transport through the membrane, different types of membrane processes can be distinguished (Table 2.1). Selective barrier structure. Transport through porous membranes is possible by viscous ﬂow or diffusion, and the selectivity is based on size exclusion (sieving mechanism). This means that permeability and selectivity are mainly inﬂuenced by membrane pore size and the (effective) size of the components of the feed: Molecules

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 2 Polymeric Membranes for Molecular Separations

20

Table 2.1 Overview of main polymer membrane characteristics

and membrane-based processes for molecular separations in liquid phase. Transmembrane gradient Selective barrier

Typical structure

Concentration difference

Pressure difference

Nonporous

anisotropic, thin-ﬁlm composite anisotropic, thin-ﬁlm composite isotropic

Pervaporation

Reverse Osmosis Nanoﬁltration

Dialysis

Nanoﬁltration

Microporous dp 2 nm Non- or microporous, with ﬁxed charge Mesoporous dp ¼ 2 . . . 50 nm Carrier in liquid

Afﬁnity ligand in solid matrix

anisotropic, isotropic track-etched immobilized in isotropic porous membrane isotropic, anisotropic

Dialysis

Dialysis

Electrical potential

Electrodialysis

Electrodialysis

Ultraﬁltration

Electroultraﬁltration

Carrier-mediated separation

with larger size than the largest membrane pore will be completely rejected, and molecules with smaller size can pass through the barrier; the Ferry–Renkin model can be used to describe the effect of hindrance by the pore on rejection in ultraﬁltration (UF) [1]. Transport through nonporous membranes is based on the solution-diffusion mechanism [1, 2]. Therefore, the interactions between the permeand and the membrane material dominate the mass transport and selectivity. Solubility and chemical afﬁnity on the one hand, and the inﬂuence of polymer structure on mobility on the other hand serve as selection criteria. However, the barrier structure may also change by uptake of substances from the feed (e.g., by plastiﬁcation), and in those cases real selectivities can be much lower than ideal ones obtained from experiments using only one component in the feed or at low feed activities. Separation using charged membranes, either nonporous (swollen gel) or porous (ﬁxed charged groups on the pore wall), is based on charge exclusion (Donnan effect; ions or molecules having the same charge as the ﬁxed ions in the membrane will be rejected, whereas species with opposite charge will be taken up by and transported through the membrane). Therefore, the kind of charge and the charge density are the most important characteristics of these membranes [1]. Finally, molecules or moieties with special afﬁnity for substances in the feed are the basis for carrier-mediated transport through the membrane; very high selectivities can be

2.2 Membrane Classification

achieved; the diffusive ﬂuxes are higher for (immobilized) liquid membranes than for polymer-based ﬁxed-carrier membranes [1]. Concentration polarization can dominate the transmembrane ﬂux in UF, and this can be described by boundary-layer models. Because the ﬂuxes through nonporous barriers are lower than in UF, polarization effects are less important in reverse osmosis (RO), nanoﬁltration (NF), pervaporation (PV), electrodialysis (ED) or carrier-mediated separation. Interactions between substances in the feed and the membrane surface (adsorption, fouling) may also signiﬁcantly inﬂuence the separation performance; fouling is especially strong with aqueous feeds. Cross-section structure. An anisotropic membrane (also called asymmetric) has a thin porous or nonporous selective barrier, supported mechanically by a much thicker porous substructure. This type of morphology reduces the effective thickness of the selective barrier, and the permeate ﬂux can be enhanced without changes in selectivity. Isotropic (symmetric) membrane cross-sections can be found for selfsupported nonporous membranes (mainly ion-exchange) and macroporous microﬁltration (MF) membranes (also often used in membrane contactors [1]). The only example for an established isotropic porous membrane for molecular separations is the case of track-etched polymer ﬁlms with pore diameters down to about 10 nm. All the above-mentioned membranes can in principle be made from one material. In contrast to such an integrally anisotropic membrane (homogeneous with respect to composition), a thin-ﬁlm composite (TFC) membrane consists of different materials for the thin selective barrier layer and the support structure. In composite membranes in general, a combination of two (or more) materials with different characteristics is used with the aim to achieve synergetic properties. Other examples besides thin-ﬁlm are pore-ﬁlled or pore surface-coated composite membranes or mixed-matrix membranes [3]. Membrane materials. Polymeric membranes are still dominating a very broad range of industrial applications. This is due to their following advantages: (i) many different types of polymeric materials are commercially available, (ii) a large variety of different selective barriers, that is, porous, nonporous, charged and afﬁnity, can be prepared by versatile and robust methods, (iii) production of large membrane area with consistent quality is possible on the technical scale at reasonable cost based on reliable manufacturing processes, and (iv) various membrane shapes (ﬂat sheet, hollow-ﬁber, capillary, tubular, capsule; Figure 2.1) and formats including membrane modules with high packing density can be produced. However, membrane polymers also have some limitations. A very well-deﬁned regular pore structure is difﬁcult to achieve, and the mechanical strength, the thermal stability and the chemical resistance (e.g., at extreme pH values or in organic solvents) are rather low for many organic polymers. In that regard, inorganic materials can offer some advantages, such as high mechanical strength, excellent thermal and chemical stabilities, and in some cases a very uniform pore shape and size (e.g., in zeolites). However, some inorganic materials are very brittle, and due to complicated preparation methods and manufacturing technology, the prices for many inorganic membranes (especially those for molecular separations) are still very high. An overview of inorganic membranes for separation and reaction processes can be found elsewhere [4, 5].

j21

j 2 Polymeric Membranes for Molecular Separations

22

Figure 2.1 Polymeric membrane shapes and cross-sectional structures. Tubular membranes are similar to flat sheet membranes because they are cast on a macroporous tube as support. Capillary membranes are hollow fibers with larger diameter, that is, >0.5 mm.

2.3 Membrane Polymer Characteristics 2.3.1 Polymer Structure and Properties

Polymers for membrane preparation can be classiﬁed into natural and synthetic ones. Polysaccharides and rubbers are important examples of natural membrane materials, but only cellulose derivatives are still used in large scale for technical membranes. By far the majority of current membranes are made from synthetic polymers (which, however, originally had been developed for many other engineering applications). Macromolecular structure is crucial for membrane barrier and other properties; main factors include the chemical structure of the chain segments, molar mass (chain length), chain ﬂexibility as well as intra- and intermolecular interactions. Macromolecule chain ﬂexibility is affected by the chemical structure of the main chain and the side groups. A macromolecule is ﬂexible when unhindered rotation around single bonds in the main chain is possible. This ﬂexibility can be reduced by several means, for example, by introducing double bonds or aromatic rings in the main chain, by forming ladder structures along the main chain or by incorporation of bulky side groups. Even larger effects with respect to the possible macroconformations can be imparted by changes of the chain architecture, that is, the transition from linear to branched or network structures. Polymer molar mass and its polydispersity have an inﬂuence on chemical and physical properties via the interactions between chain segments (of different or even the same molecule), through noncovalent binding or entanglement. For stability, high molar mass is desirable because the number of interaction sites increases with increasing chain length. However, the solubility will decrease with increasing molar mass. The preceding structural characteristics dictate the state of polymer (rubbery vs. glassy vs. semicrystalline) which will strongly affect mechanical strength, thermal stability, chemical resistance and transport properties [6]. In most polymeric membranes, the polymer is in an amorphous state. However, some polymers, especially those with ﬂexible chains of regular chemical structure (e.g., polyethylene/PE/, polypropylene/PP/or poly(vinylidene ﬂuoride)/PVDF/), tend to form crystalline

2.3 Membrane Polymer Characteristics

domains. This will lead to higher mechanical stability (high elastic modulus) as well as higher temperature and chemical resistance than for the same polymer in amorphous state, but the free volume (and hence permeability) will be much smaller. For semicrystalline polymers, the melting temperature (Tm) is important, because at this temperature a transition between crystalline and liquid state will occur. The glass transition temperature (Tg) is a much more important parameter to characterize amorphous polymers, because at this temperature a transition between solid (glass) and supercooled melt (rubber) takes place. In the glassy state molecules are frozen, therefore, chain mobility of a polymer is very limited. Heating this polymer over its Tg leads to a much more mobile and more ﬂexible state, with lower elastic modulus and higher permeability. So-called glassy polymers have a Tg beyond room temperature, and rubbery polymers (or elastomers) have a Tg below room temperature. Polymer selection will be more important for membranes with nonporous selective barrier, because ﬂux and selectivity depend on the solution-diffusion mechanism. For membranes with a porous selective barrier, the mechanical stability will be crucial to preserve the shape and size of the pores. Block- or graft copolymers, which contain two or more different repeating units within the same polymer chains, are often used instead of homopolymers in order to obtain high-performance polymeric membranes; the overriding aims are synergies between properties of the different components. In addition, blending of polymers or copolymers is also performed. In these cases, compatibility and miscibility of both (co)polymers in one solvent are required in order to get a homogenous solution (cf. Section 2.4.2). The resulting solid membrane can be a homogenous polymer blend, as indicated by one Tg value between those for the two (co)polymers. A heterogeneous (phase separated) polymer blend will be characterized by two (or more) Tg values for the individual phases. Extensive existing knowledge from polymer blending can also be adapted to membrane preparation [7]. Chemical or physical cross-linking of the polymer is applied in order to control membrane swelling, especially for separations of organic mixtures. In addition, this can also enhance mechanical strength as well as chemical stability of a membrane. However, crosslinking decreases polymer solubility, therefore it is often done after membrane formation (cf. Sections 2.4–2.6). The hydrophilicity–hydrophobicity balance of the membrane polymer is another important parameter that is mainly inﬂuenced by the functional groups of the polymer. Hydrophilic polymers have high afﬁnity to water, and therefore they are suited as a material for nonporous membranes that should have a high permeability and selectivity for water (e.g., in RO or hydrophilic PV). In addition, hydrophilic membranes have been proven to be less prone to fouling in aqueous systems than hydrophobic materials. 2.3.2 Membrane Polymer Selection 2.3.2.1 Polymers for Porous Barriers The selection of the polymer for a porous membrane is based on the requirements of the manufacturing process (mainly solubility for controlled phase separation;

j23

j 2 Polymeric Membranes for Molecular Separations

24

cf. Section 2.4.2), and the behavior and performance under application conditions. The following material properties are important to be considered: (i) Film-forming properties indicate the ability of a polymer to form a cohesive ﬁlm, and the macromolecular structure, especially molar mass and attractive interactions between chain segments, is crucial in this regard (cf. Section 2.3.1). Poly(ether sulfones) (PES), polysulfones (PSf), polyamides (PA) or polyimides (PI) are examples for excellent ﬁlm-forming materials [8]. (ii) Mechanical properties involve ﬁlm strength, ﬁlm ﬂexibility and compaction stability (especially of a porous structure). The latter is most important for high-pressure processes (e.g., for the porous substructure of an integrally anisotropic RO membrane). Because hollow ﬁber membranes are selfsupporting, the mechanical stability will be especially relevant. Many commercial ﬂat-sheet membranes are prepared on a nonwoven support material (Figure 2.2). (iii) Thermal stability requirements depend very much on the application. In order to ensure the integrity of a pore structure in the nanometer dimension, the Tg of the polymer should be higher than the process temperature. (iv) Chemical stability requirements include the resistance of the polymer at extreme pH values and other chemical conditions. Cleaning agents such as strong acids or bases, or oxidation agents are usually used to clean a fouled membrane. The stability in special solvents is also important in selected cases, that is, when processes with nonaqueous mixtures are considered. (v) The hydrophilicity–hydrophobicity balance correlates with the wettability of the material. This can be important in order to use all the pores in UF, or

Figure 2.2 SEM micrograph of a microtome cross-section of a porous polymer membrane with an anisotropic structure on a nonwoven as mechanical support (reprinted from [9], with permission from Wiley-VCH, 2006).

2.3 Membrane Polymer Characteristics

when a porous membrane is applied as a contactor between a liquid and a gas phase, and the phase boundary is stabilized because the liquid will not wet the dry pores of the membrane. With aqueous liquid phases, the ﬁrst case will require a more hydrophilic (e.g., polyacrylonitrile, PAN), and the second case a hydrophobic membrane polymer (e.g., PP). Surface wettability is also critical for fouling; cellulose is an excellent example of a hydrophilic polymer as material for low fouling UF membranes. Nevertheless, hydrophobic polymers, for instance PVDF or PES, show better chemical and thermal stability. Considering all the above criteria, PSf, PES, PAN, PVDF and cellulose-based polymers (cellulose acetates/CA/and regenerated cellulose) are mostly used for commercial UF membranes (see also Section 2.6.1). 2.3.2.2 Polymers for Nonporous Barrier The separation performance of membranes with nonporous barriers is – because of the transport via solution-diffusion (cf. Section 2.2) – predominantly inﬂuenced by the polymer material itself. Therefore, the material selection is directly related to the intrinsic (bulk) properties of the polymer, but – as for porous membranes – ﬁlmforming properties, mechanical and thermal stability form the basis of applicability (cf. Section 2.3.2.1). The following characteristics should be considered:

(i) Glassy or rubbery state of the polymer. Thermal analysis to know the Tg value is essential. The state of the nonporous polymer will determine the available free volume and the segmental mobility, and those have a decisive inﬂuence on the diffusion of molecules through the polymer. Size-based diffusion selectivity will only be possible with polymers in the rigid amorphous state. (ii) Free volume will depend on the interchain distance in the bulk of the polymer. Somewhat independent of the state (cf. above), pronounced rigidity of the main chain and very bulky side groups can lead to larger free volume and, consequently, high permeabilities. (iii) The hydrophilicity–hydrophobicity balance or other more special afﬁnities can lead to (selective) dissolution (sorption) of molecules in the membrane. When the membrane is in contact with a liquid feed, swelling can become quite large, and this effect is often dominant for selectivity (see hydrophilic vs. organophilic PV; cf. Section 2.6.3). (iv) Chemical stability requirements are similar to those for porous materials. Cleaning-related instability against active chlorine is a special problem for PA-based TFC membranes for RO. Due to the increasing number of nonaqueous applications (especially in PV and NF), polymer resistance to various organic solvents is gaining particular importance. CA, PA, PI, poly(vinyl alcohol) (PVA) and polydimethylsiloxane (PDMS) are examples of selective polymers frequently used for nonporous barriers (see also Sections 2.6.2, 2.6.3).

j25

j 2 Polymeric Membranes for Molecular Separations

26

2.3.2.3 Polymers for Charged Barrier A charged (ion-exchange) membrane is prepared from a polyelectrolyte, that is, a polymer that contains ionic side groups. An anion exchange membrane contains ﬁxed positively charged ions (e.g., NR2H þ , NR3 þ ), and this membrane will bind any anions from the feed stream. A cation exchange membrane contains ﬁxed negatively charged ions (e.g., SO3, COO), binding any cation from the feed. Exclusion of ions with the same charge depends strongly on the ﬁxed-charge density in the membrane and the electrolyte concentration outside the membrane. The basic criteria of polymer selection – ﬁlm-forming properties, mechanical and thermal stability as well as high chemical stability (extreme pH, oxidizing agents) – are similar in porous and nonporous membranes (cf. Sections 2.3.2.1 and 2.3.2.2). The following more speciﬁc important properties for ion-exchange membranes should be considered in addition:

(i) High charge density is the basis for high permselectivity. An ion-exchange membrane should be highly permeable to counterions for the ﬁxed ions, but should be impermeable to co-ions (same charge as ﬁxed ions). (ii) Low electrical resistance is achieved when the permeability of an ion-exchange membrane for the counterions with an electrical potential gradient as the driving force is high. (iii) Controlled swelling and low susceptibility to changes in external salt concentration are essential in order to keep charge density (and hence permselectivity) high, and are thus the basis for sufﬁcient stability and constant separation performance. Due to the high afﬁnity of polyelectrolytes to water, swelling is strong in ion-exchange membranes. To limit excessive swelling, chemical crosslinking is usually performed. An alternative are phase-separated polymers with ion-exchange clusters continuously distributed in a continuous hydrophobic phase. Perﬂuorosulfonic acid polymers, for example, Naﬁon, or ionic and cross-linked polystyrene derivatives, are the best known examples of ion-exchange membrane materials (see also Section 2.6.4).

2.4 Membrane Preparation 2.4.1 Track-Etching of Polymer Films

Membranes with very regular pores of sizes down to around 10 nm can be prepared by track-etching [10], and, in principle, those membranes can be used for the fractionation of macromolecules in solution. A relatively thin (<35 mm) polymer ﬁlm (typically from poly(ethylene terephthalate)/PET/or aromatic polycarbonate/PC/) is ﬁrst bombarded with ﬁssion particles from a high-energy source. These particles

2.4 Membrane Preparation

pass through the ﬁlm, breaking polymer chains and creating damaged tracks. Thereafter, the ﬁlm is immersed in an etching bath (strong acid or alkaline), so that the ﬁlm is preferentially etched along the tracks, thereby forming pores. The pore density is determined by irradiation intensity and exposure time, whereas etching time determines the pore size. The advantage of this technique is that uniform and cylindrical pores with very narrow pore-size distribution can be achieved. In order to avoid the formation of double or multiple pores, produced when two nuclear tracks are too close together, the membrane porosity is usually kept relatively low, that is, typically less than 10%. 2.4.2 Phase Separation of Polymer Solutions

Polymer membranes by phase separation. The method is often called phase inversion, but it should be described as a phase-separation process: a one-phase solution containing the membrane polymer is transformed by a precipitation/solidiﬁcation process into two separate phases (a polymer-rich solid and a polymer-lean liquid phase). Before the solidiﬁcation, usually a transition of the homogeneous liquid into two liquids (liquid–liquid demixing) occurs. The proto-membrane is formed from the solution of the membrane polymer by casting a ﬁlm on a suited substrate or by spinning through a spinneret together with a bore ﬂuid. Based on the way the polymer solution is solidiﬁed, the following techniques can be distinguished: (i) Nonsolvent-induced phase separation (NIPS) the polymer solution is immersed in a nonsolvent coagulation bath (typically water); demixing and precipitation occur due to the exchange of solvent (from polymer solution) and nonsolvent (from coagulation bath), that is, the solvent and nonsolvent must be miscible. (ii) Vapor-induced phase separation (VIPS) the polymer solution is exposed to an atmosphere containing a nonsolvent (typically water); absorption of nonsolvent causes demixing/precipitation. (iii) Evaporation-induced phase separation (EIPS) the polymer solution is made in a solvent or in a mixture of a volatile solvent and a less volatile nonsolvent, and solvent is allowed to evaporate, leading to precipitation or demixing/ precipitation. (iv) Thermally induced phase separation (TIPS) a system of polymer and solvent is used that has an upper critical solution temperature; the solution is cast or spun at high temperature, and cooling leads to demixing/precipitation. By far the majority of polymeric membranes, including UF membranes and porous supports for RO, NF or PV composite membranes, are produced via phase separation. The TIPS process is typically used to prepare membranes with a macroporous barrier, that is, for MF, or as support for liquid membranes and as gas–liquid contactors. In technical manufacturing, the NIPS process is most frequently applied, and membranes with anisotropic cross-section are obtained. Often,

j27

j 2 Polymeric Membranes for Molecular Separations

28

Figure 2.3 Schematic depiction of the continuous manufacturing process of polymeric membranes by the NIPS process.

the time before contact with the coagulation bath is used to ﬁne tune membrane pore structure; some of these processes can thus be described as combinations of VIPS followed by NIPS. Integrally anisotropic polymer membranes via NIPS process. The cross-sectional structure of an anisotropic membrane is crucial in order to combine the desired selectivity (by a barrier with pores in the lower nm range or by a nonporous polymer) with high ﬂuxes: the top layer acts as a thin selective barrier and a porous sublayer provides high mechanical strength. Such integrally asymmetric membranes were ﬁrst discovered by Loeb and Sourirajan [11]. This ﬁnding was the ﬁrst breakthrough for commercial membrane technology, that is, such RO membranes from CA showed much higher ﬂuxes than the previously produced ones from the same polymer. This method involves (Figure 2.3): (1) polymer dissolution in single or mixed solvent, (2) casting the polymer solution as a ﬁlm (proto-membrane) on suited substrate (or spinning as free liquid ﬁlm, for hollow ﬁber), (3) precipitation by immersion in a nonsolvent coagulation bath, and (4) post-treatments such as rinsing, annealing and drying. The membranes resulting from this process have typically a very thin (<1 mm, often even less than 100 nm) top skin layer (selective barrier), which is either nonporous or porous (Figure 2.4). The selection of the materials and the discussion of mechanisms for phase separation are based on ternary phase diagrams with the three main components polymer, solvent and nonsolvent; a pronounced miscibility gap (instable region) is an essential precondition. Besides thermodynamics aspects, the onset and rate of precipitation in the liquid ﬁlm (both are different depending on the distance to the plane of ﬁrst contact with the coagulation bath) are also important; the mass transfer (nonsolvent in-ﬂow, and solvent out-ﬂow) can have tremendous inﬂuence. Two mechanisms are distinguished: (i) instantaneous liquid–liquid demixing, which will result in a porous membrane, (ii) delayed onset of liquid–liquid demixing, which can result in a membrane with a nonporous barrier skin layer [2]. The rate of precipitation decreases from the top surface (in most cases, this plane of ﬁrst contact with the coagulation bath will be the barrier in the ﬁnal membrane) to the bottom surface of the cast ﬁlm. As precipitation slows down, the resulting pore sizes increase because

2.4 Membrane Preparation

Figure 2.4 SEM micrograph of a cross-section of a hollow-fiber dialysis membrane (Polyflux, Gambro) with an anisotropic structure and macrovoids in the support layer (left), and details of the inner porous separation layer in two different magnifications (right; reprinted from [12], with permission from Wiley-VCH, 2003).

the two phases have more time to separate. In practice, most systems for membrane preparation contain more than three components (e.g., polymer blends as materials and solvent mixtures for casting solution and coagulation bath). Consequently, the mechanisms can be very complex and are still under intense scientiﬁc investigation and discussion [13, 14]. Important variables to control membrane characteristics will be outlined below. Characteristics of the casting solution. Most important is the selection of a suitable solvent for the polymer, that is, the strength of mutual interactions is inversely proportional to the ease of precipitation by the nonsolvent (cf. below). Polymer concentration also plays a vital role to determine the membrane porosity. Increasing polymer concentration in the casting solution leads to a higher fraction of polymer and consequently decreases the average membrane porosity and pore size. In addition, increasing the polymer concentration could also suppress macrovoid formation and enhance the tendency to form sponge-like structure. However, this can also increase the thickness of the skin layer. Even though details depend on the properties of the membrane polymer, UF membranes can be obtained within a range of polymer concentrations of 12–20 wt%, whereas RO membranes are typically prepared from casting solutions with polymer concentrations 20 wt% (in order to increase salt rejection, a thermal annealing step is often added to the manufacturing scheme). Solvent/nonsolvent system. The solvent must be miscible with the nonsolvent (here an aqueous system). An aprotic polar solvent like N-methyl pyrrolidone (NMP), dimethyl formamide (DMF), dimethyl acetamide (DMAc) or dimethylsulfoxide (DMSO) is preferable for rapid precipitation (instantaneous demixing) upon immersion in the nonsolvent water. As a consequence, a high porosity anisotropic

j29

j 2 Polymeric Membranes for Molecular Separations

30

membrane can be achieved. For slow precipitation, yielding low porosity or nonporous membrane, solvents having a relatively low Hildebrand solubility parameter [15], like tetrahydrofuran (THF) or acetone are preferable. Additives. For certain purposes, additive or modiﬁer is added in the casting solution. Indeed, this additive can determine the performance of the ultimate membrane and is often not disclosed for commercial membranes. Usually, additives include (i) cosolvent with relatively high solubility parameter (such a solvent can slow down the precipitation rate, and higher rejection is achieved), (ii) pore-forming agents such as poly(vinyl pyrrolidone) (PVP) or poly(ethylene glycol) (PEG) (these hydrophilic additives can enhance not only membrane pore size but also membrane hydrophilicity; at least partially, these polymers form stable blends with membrane polymers such as PSf or PES), (iii) nonsolvent (should be added only in such amounts that demixing of the casting solution does not occur; promotes formation of a more porous structure and could also reduce macrovoid formation), (iv) addition of crosslinking agent into casting solution (is less frequently used, but could also reduce macrovoid formation). Characteristics of coagulation bath. The presence of a fraction of solvent in the coagulation bath can slow down the liquid–liquid demixing rate. Consequently, a less porous barrier structure should be obtained. However, the opposite effect can also occur, that is, addition of solvent can decrease polymer concentration (in the protomembrane) leading to a more open porous structure. The amount of the solvent to be added strongly depends on the solventnonsolvent interactions. As the mutual afﬁnity of solvent and nonsolvent increases, more solvent is required to achieve an effect on the membrane structure. For example, in preparation of CA membranes, the content of solvent needed in a coagulation bath for a DMSO/water system is higher than for a dioxan/water system. Instantaneous demixing resulting in a porous structure can be obtained by better miscibility between solvent and nonsolvent. In contrast, a less miscible solvent/nonsolvent combination results in a more nonporous structure. Furthermore, addition of solvent into a coagulation bath could also reduce the formation of macrovoids leading to the desired, more stable sponge-like structure of the supporting layer. Exposure time of proto-membrane before precipitation. The effect of exposure to atmosphere before immersion is dependent on the solvent property (e.g., volatility, water absorption) and atmosphere property (e.g., temperature, humidity). This step (i.e., combination of EIPS or VIPS with NIPS; cf. above) has signiﬁcant effects on the characteristics of the skin layer and the degree of anisotropy of the resulting membrane [14]. 2.4.3 Composite Membrane Preparation

Composite membranes combine two or more different materials with different characteristics to obtain optimal membrane performance. Basically, the preparation involves: (i) preparation of porous support that is usually made by a phase-separation process (cf. Section 2.4.2), and (ii) deposition of a selective barrier layer on this porous

2.4 Membrane Preparation

support. A number of methods are currently used for manufacturing asymmetric composite membranes, which will be brieﬂy discussed below [16, 17]. (i) Laminating. An ultrathin ﬁlm is cast and then laminated to a (micro)porous support. This method has been used for preparing early RO membranes for water desalination [18]. (ii) Dip-coating of a polymer solution onto a support microporous support is followed by drying, or a reactive prepolymer is applied and IR radiation is used for curing. As a result, a thin layer of the coated polymer on the substrate is obtained. In some cases, crosslinking is done during curing to increase mechanical or chemical stability. Two problems are often observed, that is, penetration of the dilute coating solution into the pores of the support and formation of defective coatings. The ﬁrst problem can be reduce by precoating the support with a protective layer from a hydrophilic polymer, such as polyacrylic acid or by ﬁlling the pores with a wetting liquid such as water or glycerin. The latter problem can be reduced by introducing an intermediate layer between the selective polymer ﬁlm and the porous substrate. (iii) Plasma polymerization. Gas-phase deposition of the barrier layer on a porous support is conducted from glow-discharge plasma via plasma polymerization. This method has been successfully used for RO membrane preparation [19]. (iv) Interfacial polymerization. This method has been developed by Cadotte et al. [20], and it is now the most important route to RO and NF membranes. The selective layer is formed in situ by polycondensation or polyaddition of reactive (bis- and trisfunctional) monomers or prepolymers on the surface of a porous support (Figure 2.5). Post-treatment such as heating is often applied in order to obtain a fully cross-linked structure of the selective barrier. Other methods derived from surface modiﬁcation, including heterogeneous graft copolymerization or in situ radical polymerization and deposition of polyelectrolyte

Figure 2.5 Schematic depiction of the preparation of TFC membranes by interfacial polymerization: The support membrane (e.g., from PES) is immersed in an aqueous monomer or prepolymer solution (e.g., bis- or trifunctional amine), and subsequently contacted with a second bath containing a water-immiscible solvent in which another reactive monomer or

prepolymer has been dissolved (e.g., bis- or trifunctional carbonic acid chloride). The reaction takes place at the interface of the two immiscible solutions on the outer surface of the support membrane, and the thickness of the polymer layer (e.g., cross-linked polyamide) is limited by its barrier properties for further diffusion of reactants into the reaction zone.

j31

j 2 Polymeric Membranes for Molecular Separations

32

layers, also gain more interest for manufacturing composite membranes with tailored selectivities (cf. Section 2.5). 2.4.4 Mixed-Matrix Membranes

Current polymeric materials are inadequate to fully meet all requirements for the various different types of membranes (cf. Section 2.2) or to exploit the new opportunities for application of membranes. Mixed-matrix membranes, comprising inorganic materials (e.g., metal oxide, zeolite, metal or carbon particles) embedded in an organic polymer matrix, have been developed to improve the performance by synergistic combinations of the properties of both components. Such improvement is either with respect to separation performance (higher selectivity or permeability) or with respect to membrane stability (mechanical, thermal or chemical). One should note that the methods to prepare such mixed-matrix membranes and the resulting properties are strongly dependent on the interactions between the different materials, and a homogeneous, regular distribution and interface compatibility are the key issues. Techniques to prepare mixed-matrix materials have been reviewed recently [21]. Mixed-matrix membranes are typically prepared by: (1) separate preparation of a polymer solution and a suspension of inorganic material, (2) mixing of both resulting in a mixed-matrix solution, (3) casting (or spinning) this solution, and (4) inducing phase separation, typically in the framework of the NIPS process (cf. Section 2.4.2). A common alternative for the preparation of mixed-matrix membranes containing inorganic oxides (e.g., silica) is the in situ synthesis of nanoparticles within a polymer solution via the sol-gel method followed by phase separation. The above techniques are mainly applied for preparation of advanced RO, NF and PV membranes (cf. Section 2.6). Only in special cases is the separation performance of the barrier really determined by the added inorganic (nano)materials, for example, by zeolites or carbon nanotubes [22].

2.5 Membrane Modification

Because most of the established membrane polymers can not meet all the performance requirements for a membrane dedicated to a particular application, membrane modiﬁcations are gaining rapidly increasing importance. Membrane modiﬁcation is aimed either to minimize undesired interactions, which reduce membrane performance (e.g., membrane fouling), or to introduce additional interactions (e.g., afﬁnity, responsive or catalytic properties) for improving the selectivity or creating an entirely novel separation function [3]. Three general approaches can be distinguished: (i) Chemical modiﬁcation of the membrane polymer (for membrane formation), (ii) Blending of the membrane polymer with (an)other polymer(s) (before membrane formation), and (iii) Surface modiﬁcation after membrane preparation.

2.5 Membrane Modification

Because the ﬁrst two approaches can involve signiﬁcant changes in composition of the casting or spinning solution, membrane structure formed during the phase separation (cf. Section 2.4.2), and, consequently, membrane properties can be quite different from the unmodiﬁed reference material. An important example of polymer modiﬁcation before membrane formation is sulfonation or carboxylation, for example, of PSf or PES, to obtain a more hydrophilic ultraﬁltration membrane from a very stable membrane polymer [23]. The most well-known example for blending with the membrane polymer is the use of the water-soluble PVP during manufacturing of ﬂatsheet or hollow-ﬁber membranes from PSf or PES [24]. Even though during the coagulation and washing steps, some of the added modiﬁed or other polymer can leach out from the membrane matrix, a fraction remains on the pore surface and thus enhances the membrane hydrophilicity. Recently, amphiphilic graft or blockcopolymers – containing functional (surface active) macromolecule segments and other segments that are compatible with the bulk of the membrane polymer – have been introduced as tailored macromolecular additives to render the ﬁnal membrane surface hydrophilic or hydrophobic [25, 26]. Surface modiﬁcation of commercially established membranes. This approach is of greatest interest in academic research but also in development within membrane companies [3, 27]. An increasing number of methods and technologies investigated for polymer surfaces in general are now being adapted to surface functionalization of polymeric membranes [28]. Highly attractive are technologies that can be integrated as another step into the continuous membrane manufacturing process (cf. Figure 2.3). A key feature of a successful (i.e., tailored) surface modiﬁcation is a synergy between the useful properties of the base membrane and the novel functional layer. In order to achieve a stable effect, chemical modiﬁcation is preferable over physical modiﬁcation. Attachment of functional moieties onto a membrane surface by physical principles can be done via the following ways [3]: (i) Adsorption/adhesion the functional layer is only physically ﬁxed on the base material, and the binding strength can be increased via multiple interactions between functional groups in the macromolecular layer and on the solid surface, (ii) Interpenetration via mixing between the added functional polymer and the base polymer in an interphase, and (iii) Mechanical interpenetration (macroscopic entanglement) of an added polymer layer and the pore structure of a membrane. In order to achieve membrane surface modiﬁcation by chemical reactions, the following approaches have been proposed [3]: (i) Heterogeneous (polymer-analogous) reactions of the membrane polymer, (ii) Grafting to (attachment of functional macromolecular moieties in one step), and (iii) Grafting from (heterogeneous graft copolymerizationof functional monomers). Photografting technologies, that is, the control of chemical surface functionalization by highly selective excitation with UV light, can be used for grafting to and grafting from and has been intensively explored for controlled functionalization of polymeric membranes [29].

j33

j 2 Polymeric Membranes for Molecular Separations

34

Fouling resistance in aqueous systems. With self-assembled monolayers, structure– property relationships for nonadsorptive and nonadhesive surfaces have been identiﬁed on a molecular level [30]. Characteristics of materials that resist the adsorption of protein should be: (a) hydrophilic/polar, (b) overall electrically neutral, (c) hydrogen-bond acceptor, and (d) not hydrogen-bond donor. In these regards, PEG, zwitterionic moieties and other materials that display kosmotropes on their surface were identiﬁed as nonfouling materials that resist protein adsorption. This principle has successfully implemented to develop fouling-resistant polymer membranes for UF and NF in aqueous systems, via membrane formation from blends with tailored graft copolymers [25] or via controlled photoinitiated grafting from [31].

2.6 Established and Novel Polymer Membranes for Molecular Separations 2.6.1 Ultrafiltration

Because the mechanisms are based on pore ﬂow and size exclusion (cf. Section 2.2), the polymer material itself does not have direct inﬂuence on ﬂux and selectivity in UF. The UF membranes usually have an integrally asymmetric structure, obtained via the NIPS technique, and the porous selective barrier (pore size and thickness ranges are 2–50 nm and 0.1–1 mm, respectively) is located at the top (skin) surface supported by a macroporous sublayer (cf. Section 2.4.2). However, the pore-size distribution in that porous barrier is typically rather broad (Figure 2.6), resulting in limited size selectivity. It should be noted that the adaptation of the NIPS or TIPS method to the fabrication of hollow-ﬁber membranes is straightforward for most systems (cf. Figure 2.4). Characterization of UF membranes is typically done via sieving

Figure 2.6 SEM micrograph of the top layer surface of an UF membrane from PAN and pore-size distribution from computerized image analysis (reprinted from [9], with permission from Wiley-VCH, 2006).

2.6 Established and Novel Polymer Membranes for Molecular Separations Table 2.2 Membrane polymers for UF and some characteristics.

Membrane polymer

Common solvent

Tg ( C)

pH range

Polysulfone Poly(ether sulfone) Poly(vinylidene ﬂuoride) Polyacrylonitrile Cellulose acetate

DMAc, DMF, DMSO, NMP DMAc, DMF, DMSO, NMP DMAc, DMF, NMP, DMSO, DMAc, DMF, nitric acid Acetone, dioxan, DMAc, DMF, DMSO, THF Stable in most organic solvents (typically prepared from cellulose acetate as precursor)

198 225 40, (Tm 175) 100 Around 135a

213 213 211 210 37

High crystalline content

49

Regenerated cellulose

a

Depending on degree of acetylation.

experiments, that is, UF of a solution of macromolecular test solutes and subsequent analysis of the changed molar mass distribution (usually by size-exclusion chromatography) are performed. The speciﬁcation of a commercial UF membrane is not the pore size, but mostly the cutoff value, that is, the molar mass for which more than 90% rejection have been observed. Isoporous track-etched membranes are a special case and typically made from PC or PET (UF-relevant pore sizes between 10 and 50 nm, and thicknesses between 8 and 25 mm; cf. Section 2.4.1). The polymers most frequently used for commercial UF membranes, along with information about their stability, are summarized in Table 2.2. Since fouling is the biggest problem for industrial application of UF, many developments in membrane materials address solutions to this problem. The following aspects should be considered in optimization of polymeric membranes for UF: (i) high permeability and rejection, (ii) hydrophilicity and fouling resistance, (iii) high maximum temperature and wide pH operating ranges, (iv) good mechanical properties, and (v) high chemical stability, especially towards cleaning agents. None of the established polymers satisﬁes all the above criteria. Chemical cross-linking during manufacturing via regeneration after NIPS of cellulose acetate can enhance cellulosic membrane performance with respect to chemical resistance and temperature stability, and a hydrophilic low-fouling UF membrane is obtained. Alternatively, surface modiﬁcations, mainly to increase hydrophilicity of membranes from hydrophobic materials, have been performed to enhance wettability and antifouling character (cf. Section 2.5). Adopted from the state-of-the-art in RO, TFC membranes have become increasingly interesting for UF as well. One of the ﬁrst examples of a commercial membrane of this type is composed of a thin barrier layer from regenerated cellulose on a porous polyoleﬁne support [32]. Signiﬁcant increase in selectivity in protein UF via electrostatic exclusion in addition to size exclusion has been achieved by introducing ﬁxed charges into the barrier layer of a cellulose-based TFC membrane [33].

j35

j 2 Polymeric Membranes for Molecular Separations

36

Similar to developments in NF and RO, solvent-resistant UF membranes could be the basis for a wide range of novel applications. Cross-linked integrally anisotropic membranes are explored with particular emphasis, and a very promising example are membranes made from poly(acrylonitrile-co-glycidyl methacrylate) that after NIPS had been cross-linked with ammonia or other tri- or difunctional amines [34]. Very uniform pore size leading to a more precise sieving would be interesting for improving UF performance. One promising approach towards this goal is pore formation via self-assembly of block-copolymers with programmed chemistry and architecture (and one crucial precondition is a low polydispersity of chain lengths). The feasibility of this approach has recently been demonstrated with the preparation of a composite membrane with 20-nm pores via formation of an ordered thin block-copolymer ﬁlm on a macroporous support membrane and subsequent selective dissolution of the polymeric pore template [35]. Very recent work indicated that such tailored block-copolymers could also be processed via NIPS to integrally anisotropic polymer membranes with rather regular pore morphology [36]. 2.6.2 Reverse Osmosis and Nanofiltration

RO and NF membranes for aqueous applications are quite similar in chemical composition and membrane preparation. However, mass transfer in NF is more complex than in RO because – in addition to solution-diffusion mechanism – size and charge exclusion are also usually involved. Ideally, polymeric membranes for RO and NF should be hydrophilic, resistant to chemicals (especially cleaning agents and chlorination) and microbial attack, and they should be structurally and mechanically stable over the long time of operation. Membranes with integrally asymmetric structure from the ﬁrst generation material CA are currently still available (e.g., for NF in common applications like water treatment). However, TFC membranes dominate in the market (e.g., FT-30, SW30, ES10/ES15 for RO and DESAL 5 and NF 270 for NF). Most of commercial RO and NF composite membranes are polyamide-based although other composite membranes, for example, with sulfonated polysulfone as selective material, are also found. Interfacial polymerization is the standard method used for preparing the PA composite structure (cf. Figure 2.5), coating is occasionally applied for other selective polymers. For RO membranes, an ultrathin nonporous polymer layer (usually crosslinked PA) is formed on the top of a highly porous membrane with very small pores (e.g., from PES or PSf ). Compared to PA-based materials, cellulosic membranes have a higher chlorine tolerance, but they are less solvent resistant and have only a narrow range of pH stability. Therefore, creating a selective material that is stable towards chlorine is still a motivation in the ﬁeld of RO. Ionexchange polymers, in particular copolymers of highly sulfonated polyethersulfone and polyethersulfone, originally developed as selective ion-conducting materials for fuel cells (cf. Sections 2.3.2.3 and 2.6.4), have recently been identiﬁed as promising candidates [37]. However, due to constraints in terms of low membrane prices and extensive process validation it would be more and more

2.6 Established and Novel Polymer Membranes for Molecular Separations

complicated to replace the established materials with such potential novel membranes. Increasing the water ﬂux considerably while keeping the selectivity high, is still a great challenge in further development of RO and NF. A very attractive TFC NF membrane with very high ﬂux at promising rejection has been prepared via layer-bylayer (LBL) deposition of polyelectrolytes on porous UF membranes [38]. A mixed matrix RO membrane, composed of a water-selective zeolite in an ultrathin PA layer, with improved permeability and unchanged high salt rejection, has been developed recently [39]. A next generation of ultrahigh ﬂux membranes could be based on composite membranes with an array of regular carbon nanotubes in the barrier layer [40]. In recent years, chemically stable membranes, which include oxidant and pHstable and (organic) solvent-resistant materials, have been intensively developed in order to broaden application of NF or RO (for a review see [41]). PI and PAN derivatives or polyether-based materials are often used for preparing chemically stable NF membranes with integrally anisotropic structure. With composite membranes (with PSf, PAN, PI or PVDF as typical support materials), the permeability and selectivity for different solvents depend strongly on the barrier polymer. Selective layers from PA, polyureas, polyphenylene oxide or sulfonated PES are more suited for polar solvents, while silicone-based layers are preferred for nonpolar solvents (in that regard, SRNF membranes can be very similar to PV membranes; cf. Section 2.6.3). Mixed-matrix membranes are often proposed to increase the membrane stability. For instance, ﬁlling PDMS with porous zeolite yielded stable SRNF membranes with enhanced ﬂuxes and selectivities allowing the use in nonpolar solvents and at high temperature [41]. A ﬁrst large-scale success example for SRNF, with integrally anisotropic polyimide membranes, is the MAXDEWAX process for crude-oil dewaxing [42]. However, to be practically useful, longterm stability and selectivity must be improved further. In addition, solvent–membrane interactions must be investigated in more detail in order to come up with satisfying and predictive models for transport and selectivity. 2.6.3 Pervaporation

As for RO and NF, most established PV membranes are composites with a nonporous polymeric barrier. In order to assure selectivity, the polymer should have preferential interactions with one of the components in the feed mixture. Integrity of the barrier is very important when separations of organic substances are concerned, and crosslinking is the preferred choice to limit swelling and improve stability. Three different types of selective barriers can be distinguished [43]: (i) hydrophilic, (ii) organophilic and (iii) organoselective ones. PAN is used as porous support for most PV membranes. This is due to its thermal stability and pronounced resistance to most organic solvents (cf. Table 2.2). Hydrophilic polymers are used as selective barriers for dehydration of organic liquids via PV. The selective layer is typically from a glassy polymer; chemically

j37

j 2 Polymeric Membranes for Molecular Separations

38

cross-linked PVA is the established material for commercial membranes [43, 44]. Furthermore, poly(acrylic acid), other polyelectrolytes, PI or chitosan have also been explored. In contrast, organophilic PV membranes are used for removal of (volatile) organic compounds from aqueous solutions. They are typically made of rubbery polymers (elastomers). Cross-linked silicone rubber (PDMS) is the state-of-the-art for the selective barrier [1, 43, 44]. Nevertheless, glassy polymers (e.g., substituted polyacetylene or poly(1-(trimethylsilyl)-1-propyne, PTMSP) were also observed to be preferentially permeable for organics from water. Polyether-polyamide block-copolymers, combining permeable hydrophilic and stabilizing hydrophobic domains within one material, are also successfully used as a selective barrier. Organoselective membranes are used for separation of organic–organic liquid mixtures. Typical applications are separations of azeotropes or mixtures of substances that have close boiling points. An example is the commercial membrane PERVAP 2256, designed for the PV separation of methanol/MTBE or ethanol/ETBE. It had been reported that the selective barrier of this membrane is most likely PVA with incorporated polar moieties [45]. The challenge in further development of polymeric PV membranes is to create materials which can increase both selectivity and permeability and have high overall stability. To control swelling, many approaches have been proposed, for example, the use of rigid-backbone polymers (e.g., PI or highly aromatic polyurea/urethane copolymers), polymer blending or chemical cross-linking. Mixed-matrix membranes may be an alternative; silica-based nanoparticles have been added to the polymer matrix to reduce swelling and increase selectivity. Filling the membrane with an organophilic adsorbent (zeolite) was also used to increase selectivity [46]. An interesting pore-ﬁlled composite membrane, made by photograft copolymerization onto a solvent-stable PAN UF membrane, has been established [47]. High ﬂux and selectivity for PV separation of organic–organic mixtures were achieved by a very thin selective barrier and prevention of swelling of the selective polymer in the pores of the barrier. Novel polymers with intrinsic microporosity (PIMs) have recently been synthesized and characterized [48]. Their highly rigid, but contorted molecular structure leads to a very inefﬁcient space ﬁlling. The polymers that are soluble in many common organic solvents form rather robust solids – including ﬂat-sheet membranes – with very high speciﬁc surface areas (600–900 m2/g). The ﬁrst examples for their use as membrane materials, indicating a promising combination of high selectivities and ﬂuxes in organoselective PV, have been reported recently. 2.6.4 Separations Using Ion-Exchange Membranes

Ion-exchange membranes are currently used not only for more or less conventional separation processes like membrane electrolysis (mainly the chlor-alkali process), electrodialysis, dialysis or electro-ultraﬁltration (cf. Table 2.1), but also in various

2.6 Established and Novel Polymer Membranes for Molecular Separations

integrated processes such as fuel cells and catalytic reactors. The preparation of ion-exchange membranes and novel developments have been reviewed in recent publications [49–52]. Two types of membranes are distinguished, namely, homogeneous and heterogeneous ones. Heterogeneous ion-exchange membranes are prepared by dispersing anion- or cation-exchange particles into a polymer matrix and subsequent extrusion of the membrane ﬁlm. Particle size signiﬁcantly inﬂuences membrane swelling as well as mechanical strength. However, currently, homogenous anion- or cationexchange membranes are preferred, composed of hydrocarbon (e.g., derivatives of styrene-divinylbenzene copolymers) or ﬂuorocarbon (e.g., Naﬁon) polymers possessing ionic groups, and supported by backing materials. Such membranes can be prepared by the following routes [50]: (i) copolymerization of a monomer containing an ion-exchange group with a nonfunctionalized monomer, (ii) modiﬁcation of a polymer ﬁlm by introducing ionic groups (e.g., by grafting from of ion-exchange polymer or of nonfunctional polymer followed by chemical functionalization), and (iii) ﬁlm casting and phase separation of solution of an ion-exchange polymer or its blend with another polymer. Route (i) is most common for styrene-divinylbenzene copolymers. Route (ii) is typically applied for hydrocarbon- (PE, PP) or ﬂuorocarbon-based membranes because it is difﬁcult to ﬁnd a suited solvent for the membrane polymers containing also highly polar ion-exchange groups. Examples of route (iii) are functionalized poly (ether ketone), polystyrene and PES; with these materials cross-linking is often performed to improve chemical stability. Overall, at moderate temperatures, ionexchange functionalized perﬂuorocarbon polymers (e.g., Naﬁon) show still the best performance. These ion-exchange membranes have hydrophobic domains (providing a nonswelling matrix) as well as polar, charged domains (providing ion-selective water channels). In general, phase-separated polymers or polymer blends seem to be superior to one-phase materials with respect to a high conductivity at not too high water sorption (swelling), which leads to nonselective passage of solutes (e.g., methanol in fuel-cell systems) [53]. Hybrid organic–inorganic materials are promising to yield membranes with high chemical and mechanical stability and excellent conductivity, and the sol-gel process in conjunction with established membrane formation is the preferred preparation method. Ion-exchange membranes with special structure and function have also been introduced. Amphoteric membranes consist of both positively (weak basic) and negatively (weak acidic) ﬁxed-charge groups, chemically bound and randomly distributed to the polymer chains; the permselectivity of these membranes is pH responsive. Charge mosaic membranes possess both cation- and anion-exchange groups arranged in oriented parallel domains separated by neutral regions, each kind of ion-exchange group provides a continuous pathway from one side of the membrane to the other [49]. Bipolar membranes that contain cation-exchange groups at one side and anion-exchange groups at the other side are interesting for several applications, such as water splitting or chemical reactions [52].

j39

j 2 Polymeric Membranes for Molecular Separations

40

2.7 Conclusion and Outlook

A wide variety of polymeric membranes with different barrier properties is already available, many of them in various formats and with various dedicated speciﬁcations. The ongoing development in the ﬁeld is very dynamic and focused on further increasing barrier selectivities (if possible at maximum transmembrane ﬂuxes) and/ or improving membrane stability in order to broaden the applicability. This tailoring of membrane performance is done via various routes; controlled macromolecular synthesis (with a focus on functional polymeric architectures), development of advanced polymer blends or mixed-matrix materials, preparation of novel composite membranes and selective surface modiﬁcation are the most important trends. Advanced functional polymer membranes such as stimuli-responsive [54] or molecularly imprinted polymer (MIP) membranes [55] are examples of the development of another dimension in that ﬁeld. On that basis, it is expected that polymeric membranes will play a major role in process intensiﬁcation in many different ﬁelds.

List of Abbreviations

CA DMAc DMF DMSO dp ED EIPS MF NF NIPS NMP PA PAN PC PDMS PE PEG PES PET PI PP PSf PV PVA

cellulose acetate dimethyl acetamide dimethyl formamide dimethyl sulfoxide pore diameter electrodialysis evaporation-induced phase separation microﬁltration nanoﬁltration nonsolvent-induced phase separation N-methyl pyrrolidone polyamide polyacrylonitrile polycarbonate polydimethylsiloxane polyethylene poly(ethylene glycol) poly(ether sulfone) poly(ethylene terephthalate) polyimide polypropylene polysulfone pervaporation poly(vinyl alcohol)

References

PVDF PVP RO SRNF TEP TFC Tg THF Tm TIPS UF VIPS

poly(vinylidene ﬂuoride) poly(vinyl pyrrolidone) reverse osmosis solvent-resistant nanoﬁltration triethyl phosphate thin-ﬁlm composite glass transition temperature tetrahydrofuran melting point temperature thermally induced phase separation ultraﬁltration vapor-induced phase separation

References 1 Baker, R.W. (2004) Membrane Technology and Applications, 2nd edn, John Wiley & Sons, Ltd, Chichester, Chapters 2, 3, 5, 6 & 9. 2 Mulder, M. (1996) Basic Principle of Membrane Technology, 2nd edn, Kluwer Academic Publishers, Dordrecht, Chapters 2, 3 & 5. 3 Ulbricht, M. (2006) Polymer, 47, 2217–2262. 4 Hsieh, P. (1996) Inorganic Membranes for Separation and Reaction, Elsevier, Amsterdam. 5 Marcano, J.G.S. and Tsotsis, T.T. (2002) Catalytic Membranes and Membrane Reactors, Wiley-VCH, Weinheim. 6 George, S.C. and Thomas, S. (2001) Progress in Polymer Science, 26, 985–1017. 7 Utracki, L.A. (2002) Polymer Blend Hand Book, Kluwer Academic Publishers, Dordrecht. 8 Ho, W.S. and Sirkar, K.K. (1992) Membrane Handbook, Van Nostrand Reinhold, New York. 9 Abetz, V., Brinkmann, T., Dijkstra, M., Ebert, K., Fritsch, D., Ohlrogge, K., Paul, D., Peinemann, K.V., Nunes, S.P., Scharnagl, N. and Schossig, M. (2006) Advanced Engineering Materials, 8, 328–358.

10 Fleischer, R.L., Alter, H.W., Furman, S.C., Price, P.B. and Walker, R.M. (1972) Science, 172, 255–263. 11 Loeb, S. and Sourirajan, S. (1962) Advances in Chemistry Series, 38, 117–132. 12 Krause, B., Storr, M., Ertl, T., Buck, R., Hildwein, H., Deppisch, R. and G€ohl, H. (2003) Chemie Ingenieur Technik, 75, 1725–1732. 13 Van de Witte, P., Dijkstra, P.J., van den Berg, J.W.A. and Feijen, J. (1996) Journal of Membrane Science, 117, 1–31. 14 Khare, V.P., Greenberg, A.R. and Krantz, W.B. (2005) Journal of Membrane Science, 258, 140–156. 15 Grulke, E.A. (1999) Polymer Handbook, 4th edn (eds J. Brandrup, E.H. Immergut, E.A. Grulke, A. Abe and D.R. Bloch), John Wiley & Sons, New York, p. VII/675, Chapter 7. 16 Cadotte, J.E. (1985) Material Science of Synthetic Membranes (ed. D.R. Lloyd), ACS Symposium Series 269, American Chemical Society, Washington DC, pp. 273–294. 17 Petersen, R.J. (1993) Journal of Membrane Science, 83, 81–150. 18 Riley, R.L., Lonsdale, H.K., Lyons, C.R. and Merten, U. (1967) Journal of Applied Polymer Science, 11, 2143–2158.

j41

j 2 Polymeric Membranes for Molecular Separations

42

19 Yasuda, H. (1984) Journal of Membrane Science, 18, 273–284. 20 Cadotte, J., King, R., Majerle, R. and Petersen, R. (1981) Journal of Macromolecular Science-Chemistry, A15, 727–755. 21 Kickelbick, G. (2003) Progress in Polymer Science, 28, 83–114. 22 Sholl, D.S. and Johnson, J.K. (2006) Science, 312, 1003–1004. 23 M€ ockel, D., Staude, E. and Guiver, M.D. (1999) Journal of Membrane Science, 158, 63–75. 24 Boom, R.M., van den Boomgaard, Th. and Smolders, C.A. (1994) Journal of Membrane Science, 90, 231–249. 25 Asatekin, A., Menniti, A., Kang, S., Elimelech, M., Morgenroth, E. and Mayes, A.M. (2006) Journal of Membrane Science, 285, 81–89. 26 Rana, D., Matsuura, T. and Narbaitz, R.M. (2006) Journal of Membrane Science, 277, 177–185. 27 Nunes, S.P. and Peinemann, K.V. (2006) Membrane Technology in the Chemical Industry, 2nd edn, (eds S.P. Nunes and K.V. Peinemann), Wiley-VCH, Weinheim, p. 1, Part 1. 28 Kato, K., Uchida, E., Kang, E.T., Uyama, Y. and Ikada, Y. (2003) Progress in Polymer Science, 28, 209–259. 29 He, D.M., Susanto, H. and Ulbricht, M. (2009) Progress in Polymer Science, 34, 62–98. 30 Kane, R.S., Deschatelets, P. and Whitesides, G.M. (2003) Langmuir, 19, 2388–2391. 31 Susanto, H. and Ulbricht, M. (2007) Langmuir, 23, 7818–7830. 32 Millipore Corp.; technical publications: http://www.millipore.com/ techpublications/tech1/5f5nrn. 33 van Reis, R., Brake, J.M., Charkoudian, J., Burns, D.B. and Zydney, A.L. (1999) Journal of Membrane Science, 159, 133–142. 34 Hicke, H.G., Lehmann, I., Malsch, G., Ulbricht, M. and Becker, M. (2002) Journal of Membrane Science, 198, 187–196.

35 Yang, S.Y., Ryu, H., Kim, H.Y., Kim, J.K., Jang, S.K. and Russell, T.P. (2006) Advanced Materials, 18, 709–712. 36 Peinemann, K.V., Abetz, V. and Simon, P.F.W. (2007) Nature Materials, 6, 992–996. 37 Park, H.B., Freeman, B.D., Zhang, Z.B., Sankir, M., and McGrath, J.E. (2008) Angew. Chem. Int. Ed., 47, 6019–6024. 38 Malaisamy, R. and Bruening, M.L. (2007) Langmuir, 21, 10587–10592. 39 Jeong, B.H., Hoek, E.M.V., Yan, Y., Subramani, A., Huang, X., Hurwitz, G., Ghosh, A.K. and Jawor, A. (2007) Journal of Membrane Science, 294, 1–7. 40 Majumder, M., Chopra, N., Andrews, R. and Hinds, B.J. (2005) Nature, 438, 44–44. 41 Vandezande, P., Gevers, L.E.M. and Vankelecom, I.F.J. (2008) Chemical Society Reviews, 37, 365–405. 42 White, L.S. and Nitsch, A.R. (2000) Journal of Membrane Science, 179, 267–274. 43 Jonquieres, A., Clement, R., Lochon, P., Neel, J., Dresch, M. and Chretien, B. (2002) Journal of Membrane Science, 206, 87–117. 44 Feng, X. and Huang, R.Y.M. (1997) Industrial & Engineering Chemistry Research, 36, 1048–1066. 45 Sharma, A., Thamphi, S.P., Suggala, S.V. and Bhattacharya, P.K. (2004) Langmuir, 20, 4708–4714. 46 Vankelecom, I.F.J., Depre, D., De Beukelaer, S. and Uytterhoeven, J.B. (1995) The Journal of Physical Chemistry, 99, 13193–13197. 47 Ulbricht, M. and Schwarz, H.H. (1997) Journal of Membrane Science, 136, 25–33. 48 McKeown, N.B. and Budd, P.M. (2006) Chemical Society Reviews, 35, 675–683. 49 Strathmann, H. (2004) Ion Exchange Membrane Separation Processes, Elsevier, Amsterdam. 50 Xu, T. (2005) Journal of Membrane Science, 263, 1–29. 51 Nasef, M.M. and Hegazy, E.S.A. (2004) Progress in Polymer Science, 29, 499–561.

References 52 Balster, J., Stamatialis, D.F. and Wessling, M. (2004) Chemical Engineering Progress, 43, 1115–1127. 53 Robeson, L.M., Hwu, H.H. and McGrath, J.E. (2007) Journal of Membrane Science, 302, 70–77.

54 Geismann, C., Yaroshchuk, A. and Ulbricht, M. (2007) Langmuir, 23, 76–83. 55 Ulbricht, M. (2004) Journal of Chromatography B, 804, 113–125.

j43

j45

3 Fundamentals of Membrane Solvent Separation and Pervaporation Bart Van der Bruggen

3.1 Introduction: Separation Needs for Organic Solvents

Separation processes for product recovery and puriﬁcation represent more than 40% of the energy needs in chemical production processes. These include removal of impurities from raw materials, separation of products and by-products after reaction, and separation of pollutants from water and process streams. Classical solutions to the former two are based on thermodynamic equilibria involving a phase transition; distillation and liquid–liquid extraction are typical examples [1]. All these processes use either energy (e.g., distillation) or mass (e.g., liquid–liquid extraction) as the separating agent, which is usually not sustainable in terms of energy consumption and/or waste generation. Membrane separations are a totally different class of processes. Although some may argue that a membrane can be considered a mass-separating agent, transport properties of compounds that have to be separated always determine the process efﬁciency, not the equilibrium between two contacting phases. This makes membrane processes a new and different class of separations. Thermodynamic calculations conﬁrm that membrane separations have intrinsically a substantially lower energy consumption and a higher exergetic efﬁciency [2–4]. This requires the use of membranes as separating tools similar to the operating methods of classical processes. Nevertheless, membranes are still mainly used in areas where the proﬁts that can be obtained (both the economical proﬁts and the environmental proﬁts) are relatively small, compared to the challenges in the process industry. Membranes are widely used in water treatment, for drinking-water production [5, 6], desalination [7], wastewater treatment [8, 9] and process water recycling [10–12]. The economical and environmental beneﬁts are obvious for these applications. However, in these applications membranes are used only for puriﬁcation, that is, to remove solutes from the (water) matrix. The desired purity determines which membrane is to be used. In the context of increasing demands for purity – drinking water free of any anthropogenic compound, wastewater puriﬁed to be reused – it is understandable that most attention is devoted to those pressure-driven ﬁltration processes delivering

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

46

a high-purity product (reverse osmosis and, to a certain extent, nanoﬁltration). Energy consumption is the main parameter to be optimized, low-pressure reverse-osmosis membranes or nanoﬁltration membranes with low cutoff being the most interesting compromise. In this context, nanoﬁltration is an exception since it is the only process where separation is achieved on purpose, not as a side effect: nanoﬁltration membranes should allow passage of monovalent salts and retain multivalent salts. Separations between liquid matrix compounds is more challenging and less developed, in spite of the large beneﬁts that can be (theoretically) attained. Environmental beneﬁts of solvent separations are mainly in a drastic reduction of energy consumption, although considerable effects on wastewater generation can be obtained as well. The main reference process to achieve this is pervaporation [13]. This is not a new process, but it had for a long time difﬁculties in ﬁnding its position. Pervaporation is not the most suitable process for puriﬁcation, but it is a powerful and underestimated tool for separation between solvents. Separation factors in pervaporation can be 100–200, whereas the separation factors between mono- and divalent salts in nanoﬁltration are usually 4–6 and never above 10. For organic solutes in solvents, both processes are surprisingly similar in terms of separation and transport, in spite of different operating principles. This will be made clear from a discussion of membrane materials and properties. The translation of operation principles to ﬂux and separation prediction in both cases will also be discussed.

3.2 Pervaporation and Nanofiltration Principles

Pervaporation is a concentration-driven membrane process for liquid feeds. It is based on selective sorption of feed compounds into the membrane phase, as a result of differences in membrane–solvent compatibility, often referred to as solubility in the membrane matrix. The concentration difference (or, in fact, the difference in chemical potential) is obtained by applying a vacuum at the permeate side, so that transport through the membrane matrix occurs by diffusion in a transition from liquid to vapor conditions (Figure 3.1). Alternatively, a sweep gas can be used to obtain low vapor pressures at the permeate side with the same effect of a chemical potential gradient. The precise point of transition is undetermined; most theories assume that this would occur inside the membrane during transport. This leads to a ﬁve-step

Figure 3.1 Schematic diagram of the pervaporation process.

3.2 Pervaporation and Nanofiltration Principles

description: sorption – liquid diffusion – vaporization – vapor diffusion – desorption. The physical relevance of this description may be questioned since solvent molecules permeating through the membrane are not a continuous phase but rather appear as individual molecules, or as small groups of molecules. Therefore, it is incorrect to consider the permeating substances as a liquid or a vapor; they are rather sorbed molecules. Separation is determined by factors affecting sorption (polarity, hydrophilicity/hydrophobicity differences, but also size) and molecular diffusion (sterical hindrance, interaction effects, dragging, competition). Therefore, typical separations in pervaporation are those where large differences in these parameters are present, mainly separations between water and an organic solvent, the latter preferably large and nonpolar. Hydrophilic pervaporation membranes are used for dehydration of organic solvents; in this application, the combination of a solvent and a polymer requires an optimal chemical stability. Hydrophobic membranes are used for the removal of (small) organic contaminants from water; these compounds can also be aggressive towards the membrane, in particular when present in relatively high concentrations. Because the driving force in pervaporation is a partial vapor pressure gradient, the process is only economical when the concentrations of contaminants to be removed are sufﬁciently high (no trace impurities) or when their driving force is high enough (possibly by operating at higher temperature), which again increases the need for membrane stability. Solvent-resistant nanoﬁltration is based on a pressure gradient as the driving force. Pressure affects the chemical potential, so that the driving force again translates to a difference in chemical potential between feed and permeate. Solution-diffusion, similar to reverse osmosis and pervaporation, is often proposed as the determining transport mechanism [14]. This implies that nearly the same parameters are of importance as for pervaporation. Nevertheless, nanoﬁltration membranes may have a larger free space available for transport – some may denote this free space as pores, but this is probably incorrect, although there is no strict deﬁnition of a pore; nanovoids is the scientiﬁcally correct description. When nanovoids become larger, they come close to the micropores observed in ultraﬁltration; in this case, transport may be rather determined by viscous ﬂow. Interactions with the membrane material are less intense here, and molar size remains as the only parameter determining separation. Thus, (solvent resistant) nanoﬁltration is related to both pervaporation and ultraﬁltration. Molar size emerges as the parameter to be used as the main discriminating factor in this case, although it is evident that other parameters will also play a role for membranes with a low cutoff. When organic solutes are considered, nanoﬁltration membranes are usually described by a single cutoff value, reﬂecting the large inﬂuence of molar size, but ignoring other interactions. A major problem that has been identiﬁed is the inﬂuence of the solvent itself on swelling, pore solvation and solute solvation [15], leading to different cutoff values and ﬂuxes depending on the solvent used [16, 17]. Applications are to be found in separation of relatively small organic solutes, in the range of 300–1000 g/mole, from any organic solvent [18–20]. Large-scale applications have been in operation since 1998 [21], the best known being the MAX-DEWAX process.

j47

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

48

3.3 Membrane Materials and Properties for Solvent Separation 3.3.1 Solvent-Stable Polymeric Membrane Materials

Traditional polymers for separations in water have a limited chemical resistance and are not useful for solvent separations. Some may be applicable in nonaggressive solvents such as methanol and ethanol due to crosslinking, additives or additional interlayers, but not in any other solvent; modiﬁed polyamide membranes and poly (ethersulfone) membranes are typical examples. Pervaporation membranes have always been intended for use in demanding conditions, including application in organic solvents. Therefore, a wide range of solvent stable membranes has been studied, usually with asymmetric structure, a dense top layer and several porous sublayers. Pervaporation membranes are prepared by dip-coating, plasma polymerization or interfacial polymerization [22]. Swelling can be a problem, because of the intense contact between liquid feed and the membrane top layer. The permeate side of the membrane is under vacuum and therefore not swollen. This results in an asymmetric structure that may cause stability problems. Nevertheless, swelling is needed to some extent in view of obtaining high ﬂuxes, so that amorphous polymers with a sorption value of 5–25% by weight are to be preferred. Membrane thickness ranges from 100 nm to several mm, depending on how defect free the top layer can be made; often a somewhat thicker structure is preferred to avoid defects. A list of typical commercial pervaporation membranes [23] is given in Table 3.1. Commercial hydrophilic membranes are very often made of polyvinyl alcohol (PVA), with differences in the degree of crosslinking. Commercial hydrophobic membranes often have a top layer in polydimethyl siloxane (PDMS). However, a wide variety of membrane materials for pervaporation can be found in the literature, including polymethylglutamate, polyacrylonitrile, polytetraﬂuoroethylene, polyvinylpyrrolidone, styrene-butadiene rubber, polyacrylic acid, and many others [24]. A comprehensive overview of membrane materials for pervaporation is given by Semenova et al. [25]. Solvent resistant nanoﬁltration membranes are a much more recent evolution. Historically, the membranes developed by Membrane Products Kyriat Weizmann (Israel) – now Koch – (MPF 44, MPF 50, MPF 60) were the ﬁrst nanoﬁltration membranes intended for application in organic solvents, although other membranes (e.g., PES and PA membranes) also have a limited solvent stability. The Koch membranes are based on PDMS, similarly to pervaporation membranes, although the level of crosslinking is quite different. Other membrane materials include mainly polyimide, polyacrylonitrile and polybenzimidazole. An overview of commercially available membranes is given in Table 3.2. These membranes are manufactured in procedures usually derived from practical experience; by using high-throughput screening, it was shown that optimization is possible [26]. Many other membrane materials are described in the scientiﬁc literature and in patents; an overview is given by Cuperus and Ebert [27].

3.3 Membrane Materials and Properties for Solvent Separation Table 3.1 Pervaporation membranes used in commercial applications [23].

Brand name

Manufacturer

Material

PERVAP 2200 PERVAP 2201 PERVAP 2202 PERVAP 2205 PERVAP 2210 PERVAP 2510 CM-Celfa GKSS Simplex PERVAP 1060 PERVAP 1070 MTR 100 MTR 200

Sulzer Chemtecha Sulzer Chemtecha Sulzer Chemtecha Sulzer Chemtecha Sulzer Chemtecha Sulzer Chemtecha CM-Celfab GKSSc Sulzer Chemtecha Sulzer Chemtecha MTRd MTRd

GKSS PEBA GKSS PDMS GKSS PMOS

GKSSc GKSSc GKSSc

PVA cross-linked/PAN support PVA highly cross-linked/PAN PVA specially cross-linked/PAN PVA specially cross-linked/PAN PVA lightly cross-linked/PAN PVA specially cross-linked/PAN PVA cross-linked/PAN Complex polyelectrolytes/PAN PDMS cross-linked/PAN support PDMS cross-linked þ silicalite/PAN PDMS cross-linked/porous support EPDM/PDMS cross-linked/ porous support PEBA/porous support PDMS cross-linked/porous support PMOS cross-linked/porous support

Hydrophilic/ hydrophobic Hydrophilic Hydrophilic Hydrophilic Hydrophilic Hydrophilic Hydrophilic Hydrophilic Hydrophilic Hydrophobic Hydrophobic Hydrophobic Hydrophobic Hydrophobic Hydrophobic Hydrophobic

a

Sulzer Chemtech Ltd, Winterthur, Switzerland. CM-Celfa, Seewen-Schwyz, Switzerland. c GKSS, Geesthacht, Germany. d Membrane Technology and Research Inc., Menlo Park, CA. b

Among the various materials are crosslinked PAN, polyphosphazenes, polyphenylenesulﬁde, polyetheretherketone, and various polymer blends [28–31]. Particularly interesting is the use of zeolites as ﬁller in organic polymers, which aims at improving the performance of (silicone-based) membranes for separations in nonpolar solvents, by adding more cross-links to the membrane material [32, 33]. 3.3.2 Ceramic Membrane Materials

Ceramic membranes may overcome some of the disadvantages of polymeric membranes, particularly the chemical resistance. The higher cost of ceramic membranes may be compensated by the signiﬁcantly higher ﬂuxes, especially at high temperatures. A full comparison of polymeric and ceramic membranes is given in Table 3.3. Some efforts have already been made to develop ceramic pervaporation membranes, especially silica and zeolite membranes, which are both hydrophilic membranes. Silica pervaporation membranes have been developed by ECN, The Netherlands. The membranes were tested in a pilot installation of 1 m2 membrane surface at Akzo Nobel and other companies in the Netherlands [34, 35]. A-type zeolite pervaporation membranes have been developed by Mitsui Engineering & Shipbuilding Co Ltd, which have been implemented in an industrial

j49

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

50

Table 3.2 Commercial solvent-resistant nanofiltration

membranes with characteristics as specified by the manufacturers.

Membrane

Manufacturer

Material

MWCO (Da)

Tmax ( C)

L (l/h m2 bar)

R (%)

N30F NF-PES-010 MPF-44 MPF-50 Desal-5-DK Desal-5-DL SS-030505 SS-169 SS-01 StarMem-120 StarMem-122 StarMem228

Nadira Nadira Kochb Kochb Osmonicsc Osmonicsc SolSepd SolSepd SolSepd MET e MET e MET e

PES PES PDMS PDMS PA PA n.k. n.k. n.k. PI PI PI

400 1000 250 700 150–300 150–300 — — — 200 220 280

95 95 40 40 90 90 90 150 150 60 60 60

1.0–1.8h 5–10h 1.3h 1.0i 5.4h 9.0h 1.0 j 10 j 10 j 1.0k 1.0k 0.26k

70–90l 30–50l 98m — 98n 96n >90p 95p 97q — — —

a

Nadir Filtration GmbH, Wiesbaden, Germany; Koch Membrane Systems, Wilmington, MA, USA; c GE Osmonics, Vista, CA, USA; d SolSep BV, Apeldoorn, The Netherlands; g Membrane Extraction Technology, London, UK; h pure-water permeability; i methanol permeability; j ethanol permeability; k toluene permeability; l 4% lactose (MW 342); m 5% sucrose (MW 342); n MgSO4; p MW500 in ethanol; q MW1000 in acetone; n.k. not known; — not speciﬁed. b

Table 3.3 Comparison of polymeric and ceramic pervaporation membranes.

Polymeric membrane materials

Ceramic membrane materials

Low production cost Production upscaling easy Variation in module form easy Stability at long term unknown Limited versatility in organics Vulnerable for unknown components in mixtures Thermal regeneration impossible High-temperature applications impossible

High production cost Production upscaling difﬁcult Variation in module form difﬁcult Stability at long term expected good Good versatility in organics Resistance to unknown component expected good Thermal regeneration possible High-temperature applications possible

3.3 Membrane Materials and Properties for Solvent Separation

dehydration plant in Japan [36]. Pervaporation membranes with an active layer of zeolite and amorphous silica on porous supports of alumina or stainless steel have become commercially available [37]; research on other zeolite-type materials (e.g., silicalite, ZSM-5, T-type zeolite) is going on, but so far only at the laboratory scale. Ceramic membranes might also be a signiﬁcant improvement in solvent-resistant nanoﬁltration, although the cost of ceramic membranes is relatively high. To date, only a few ceramic membrane types are commercially available, in spite of the good performance of these membranes. Hydrophilic ceramic nanoﬁltration membranes in asymmetric multilayer conﬁgurations have been successfully developed since the late 1990s [38–40]. These consist of an open porous support, mesoporous interlayers, and defectless microporous top layers. Support layers may combine an extruded body containing coarse pores in the mm range, and a slip-casted layer on top with pores of the order of 50–100 nm. The most often used material is a-alumina, although titania is tending to become more popular due to its higher chemical stability [41]. The intermediate layers (usually more than one to prevent defects) are used to gradually decrease the pore size and the surface roughness of the membrane. The interlayers are prepared by the colloidal sol-gel procedure [42], often using alumina in the g-alumina phase, or titania and even silica or zirconia. g-alumina interlayers are relatively thick (about 2 mm) while titania or zirconia interlayers are thinner (about 0.5 mm). Therefore, g-alumina interlayers are more suitable to cover irregularities and defects. The obvious disadvantage is that g-alumina layers lead to lower solvent ﬂuxes than titania layers, because of their higher ﬂuid resistance. Furthermore, g-alumina is chemically unstable in acid solutions (pH < 3) and in alkaline solutions (pH > 11). The pore size of the interlayers depends on the hydrolyzing/peptisizing process and the calcination temperature and should be about 3–5 nm to obtain nanoﬁltration membranes. Calcination temperatures should be below the phasetransition temperature of the used metal oxide; for titania this involves the anatase–rutile transition in the temperature region 500–700 C [40]. The top layer contains the smallest pores and deﬁnes the membranes nanoselectivity. Top layers can be made of alumina (boehmite), titania, zirconia, silica, or mixtures of these. Pore diameters are in the order of 1 nm and are obtained by applying the polymeric sol-gel method [42]. The calination temperature applied in this procedure should be low enough in order to avoid sintering effects, and consequent pore growth. The calcination temperature does not only determine the ﬁnal pore size of the top layer, but also the phase structure of the top-layer material. For example, when the calcination temperature for a titania top layer is 200 C, titania is amorphous with very small pores (1–2 nm), from 300 C onwards titania is in an anatase phase having pores in the range of 2–4 nm. Amorphous titania, however, is less resistant to corrosion, and therefore has a smaller applicable pH range [40]. Metal oxides, used for manufacturing of ceramic nanoﬁltration membranes, are intrinsically hydrophilic. This limits the use of these membranes to polar solvents; ﬁltration of nonpolar solvents (n-hexane, toluene, cyclohexane) usually yields zero ﬂuxes. Attempts have been made to modify the pore structure by adding hydrophobic groups, for example, in a silane coupling reaction [38, 43]. This approach is similar to modiﬁcations of ultraﬁltration and microﬁltration membranes

j51

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

52

using chloroalkylsilanes [44–48], phosphonic acids [49, 50] or ﬂuoroalkylsilanes [51], where the effect is twofold, that is, a combination of pore-size control and tuning of hydrophobicity. However, the precise interaction between the chlorosilanes and the membrane surface is very complex and there is in many cases no evidence that it involves a chemical reaction rather than an adsorption reaction [52, 53]. Due to the lack of a real chemical reaction between organochlorosilanes and the membrane surface, the stability of the modiﬁcation with these reactants may be limited, and to date no commercial hydrophobic ceramic nanoﬁltration membranes have been developed. 3.3.3 Solvent Stability

A difﬁcult problem that prevented the use of nanoﬁltration in organic solvents for a long time was the limited solvent stability of polymeric nanoﬁltration membranes, and the lack of ceramic nanoﬁltration membranes. For polymeric membranes, different problems occurred: zero ﬂux due to membrane collapse [54], inﬁnite nonselective ﬂux due to membrane swelling [54], membrane deterioration [55], poor separation quality [56], etc. In an early study of four membranes thought to be solvent stable (N30F, NF-PES-10, MPF 44 and MPF 50), it was observed that three of these showed visible defects after ten days exposure to one or more organic solvents, and the characteristics of all four membranes changed notably after exposure to the solvents [15]. This implies that these membranes should be denoted as semi-solventstable instead of solvent stable. Less information is available about the stability of ceramic membranes. It is generally thought that ceramic membranes have excellent solvent stability. Acid conditions may be more problematic; it was shown [57] that an alumina nanoﬁltration membrane was very sensitive to corrosion effects in dynamic experiments, whereas the performance of a similar titania membrane was stable in the pH range from 1.5 to 13. 3.3.4 Structural Properties for Membranes in NF and PV

Due to recent advances in membrane development, nanoﬁltration membranes are nowadays increasingly used for applications in organic solvents [27, 58]. This narrows the gap between pervaporation and nanoﬁltration. It is even possible that the requirements for membrane structures completely overlap for the two processes: whereas membrane stability becomes more important for nanoﬁltration membranes, the performance of pervaporation membranes could be improved by using an optimized (thinner) structure for the top layers. It might even be possible to use the same membranes in both applications. At this moment it is not possible to deﬁne which membrane structure is necessary for nanoﬁltration or for pervaporation, and which membrane is expected to have a good performance in nanoﬁltration, in pervaporation or in both. Whereas pervaporation membranes are dense, nanoﬁltration membranes

3.4 Flux and Separation Prediction

can be either dense or porous. For this reason, some nanoﬁltration and pervaporation membranes might be interchangeable. This was suggested for polymeric membranes [59]. Khayet and Matsuura [60] explored a similar relation between pervaporation and membrane distillation using polyvinylidene ﬂuoride membranes. Similar trends are developing for ceramic membranes applied in pervaporation and nanoﬁltration, although much slower because ceramic pervaporation and nanoﬁltration membranes are still sparsely available; more experimental observations and experience with applications are needed in this ﬁeld. Promising results were obtained by Sekulic et al. [61] for titania membranes that can be used in pervaporation as well as nanoﬁltration.

3.4 Flux and Separation Prediction 3.4.1 Flux Models in NF

For relatively porous nanoﬁltration membranes, simple pore ﬂow models based on convective ﬂow will be adapted to incorporate the inﬂuence of the parameters mentioned above. The Hagen–Poiseuille model and the Jonsson and Boesen model, which are commonly used for aqueous systems permeating through porous media, such as microﬁltration and ultraﬁltration membranes, take no interaction parameters into account, and the viscosity as the only solvent parameter. It is expected that these equations will be insufﬁcient to describe the performance of solvent resistant nanoﬁltration membranes. Machado et al. [62] developed a resistance-in-series model based on convective transport of the solvent for the permeation of pure solvents and solvent mixtures: J¼

DP f ½ðg c g l Þ þ f1 h þ f2 h 0

where f1 and f2 are solvent independent parameters characterizing the nanoﬁltration and ultraﬁltration sublayers, f0 a solvent parameter, g c the critical surface tension of the membrane material and g l the surface tension of the solvent. This model is also based on the dependence of the ﬂux on two parameters, namely the solvent viscosity and the difference in surface tension between the solid membrane material and the liquid solvent. However, this model does not cover the whole area of membranes and solvents, as shown by Yang et al. [63]. The model is developed for hydrophobic membranes, but seems inadequate for the description of ﬂuxes through hydrophilic membranes. Moreover, for each solvent–membrane combination an empirical parameter f0 has to be determined as a measure for the interaction between a solvent and the membrane material. It will be attempted to replace this parameter by a combination of nonempirical parameters. Polymeric membranes with a less porous structure, pervaporation membranes as well as nanoﬁltration membranes, can be described by a solution-diffusion mecha-

j53

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

54

nism, possibly corrected for the inﬂuence of convective transport [64]. A description of solvent transport in this case is necessarily based on the solution-diffusion (SD) model [65]. With respect to ﬂux modeling of organic solvents, a diffusion based model was presented by Bhanushali et al. [66]: Vm 1 J/ h fn g m This model combines different approaches of existing models by introducing at the same time the solvent viscosity, the molar volume Vm (as a measure for the molecular size), the surface tension of the solid membrane material and a sorption value f (as a measure for membrane–solvent interactions). Other SD-based transport models were presented by White [14], providing a predictive model for feed solutions with a high concentration of aromatics, by Scarpello et al. [67] and by Gibbins et al. [68]. A slightly modiﬁed equation was proposed by Geens et al. [69]: J/

Vm h Dg

where Dg is the difference in surface tension (mN/m), h is the dynamic viscosity (Pa s), and Vm is the solvent molar volume (m3/mol). Transport models for the description of solute transport in aqueous solution are the Spiegler–Kedem model and the solution-diffusion model [65]. The former model incorporates both viscous and diffusive ﬂow, whereas the latter can only be used for transport through dense membranes by solute diffusion. White [14] presented an SD-based model for the permeation of several reference solutes, dissolved in toluene, through dense membranes. Bhanushali et al. [66] succeeded in describing experimental data with the Spiegler–Kedem model. Gevers et al. [70] and Vankelecom et al. [71] used the reformulated solution-diffusion model of Paul and the Kedem–Katchalsky model to explain solute ﬂuxes; it was shown that solutes with a high molar volume were most inﬂuenced by diffusive transport, whereas solutes with a low molar volume are dominantly transported by convection. Matsuura and Sourirajan [72] developed a model for convective transport of dissolved components, incorporating a solvent-dependent pore diameter. Gibbins et al. [70] calculated the pore radii of MPF-50 and Desal-5-DK based on ﬁltration experiments carried out in methanol, using several models for convective ﬂow through porous membranes. The different attempts for the modeling of nonaqueous solute transport provide, however, models that are limited to speciﬁc experimental data. A new approach is the application of chemometrics (and neural networks) in modeling [73]. This should allow identiﬁcation of the parameters of inﬂuence in solvent-resistant nanoﬁltration, which may help in further development of equations. Development of a more systematic model for description and prediction of solute transport in nonaqueous nanoﬁltration, which is applicable on a wide range of membranes, solvents and solutes, is the next step to be taken. The Maxwell–Stefan approach [74] is one of the most direct methods to attain this.

3.4 Flux and Separation Prediction

3.4.2 Rejection in NF

Similar to the approach for solvents, both diffusive and convective transport of solutes can be modeled separately. For dense membranes, a solution-diffusion model can be used [14], where the ﬂux Ji of a solute is calculated as: ðVi ðPf Pp ÞÞ cf ;i cp;i exp RT Ji ¼ Di Ki Dx with Di the diffusivity of the solvent in the polymer matrix, Ki the partition coefﬁcient between component i and polymer, cf,i and cp,i the concentration of component i in the feed or permeate (mol/l), Vc the molar volume of component c at the boiling point (m3/mol), Pf the feed side pressure (bar), Pp the permeate-side pressure (bar), R the universal gas constant (J/mol K), T the absolute temperature (K), and Dx the membrane thickness (m). This equation yields a good description of solute transport, but it is not possible to predict separations because diffusivities and partition coefﬁcients have to be related to measurable membrane/solvent parameters. The transport equations of Spiegler and Kedem combine both diffusion and convection: Js ¼ LðDPsDpÞ Jc ¼ Ps Dx

dc þ ð1sÞJs c dx

The rejection of a given molecule can then be calculated as: sð1FÞ 1s with F ¼ exp R¼ Js 1sF Ps The permeability Ps is a measure of the transport of a molecule by diffusion. The reﬂection coefﬁcient s of a given component is the maximal possible rejection for that component (at inﬁnite solvent ﬂux). Various models have been proposed for the reﬂection coefﬁcient [75–77]. In the lognormal model [78], a lognormal distribution is assumed for the pore size. No steric hindrance in the pores or hydrodynamic lag is taken into account, but it is assumed that a molecule permeates through every pore that is larger than the diameter of the molecule. Moreover, the diffusion contribution to the transport through the membrane is considered to be negligible. Therefore, the reﬂection curve can be expressed as: ! 1 1 ðlnðrÞlnðr ÞÞ2 pﬃﬃﬃﬃﬃﬃ exp dr s¼ 2S2p Sp 2p r ðrc 0

with rc ¼ dc/2. This equation comprises two variables, Sp and r , where Sp is the standard deviation of the distribution. This standard deviation is a measure for the

j55

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

56

Figure 3.2 Pore solvation solute solvation may influence the rejection of solutes in organic solvents [80, 81].

distribution of the pore sizes. r is a mean pore size, namely the size of a molecule that is retained for 50%. However, it is clear that other parameters than only solute size determine transport and rejection. Tarleton et al. [17, 79] showed that polarity has a major inﬂuence on permeation. Geens et al. [80, 81] showed that interaction effects between solvents, solutes and the membrane material determine the rejection of a given solute (Figure 3.2). Differences in solvation may result in lower rejections; this explains the differences in molecular weight cutoff that were observed. These effects were observed for both polymeric and ceramic NF-membranes, showing that differences in rejection are not due to swelling. Nevertheless, swelling effects have been demonstrated by Tarleton et al. [82, 83] and are known to affect transport in polymeric membranes. 3.4.3 Models for PV: from Solution-Diffusion to Maxwell–Stefan

The transport mechanism in polymeric pervaporation is generally understood as a combined sorption–diffusion–desorption process. Simple sorption–diffusion models [84] can serve as a starting point for modeling the membrane separation. These models give the ﬂux of a component through the membrane as a function of concentration or partial pressure differences over the membrane. This approach, however, does not incorporate coupling and interaction effects that are possible between the different components in a mixture. According to Lipnizki et al. [85], coupling effects can be expected during all three stages in the pervaporation process (sorption, diffusion, desorption). Current research focuses on thermodynamic models to describe sorption of different components into the membrane, and on the Maxwell–Stefan formulation for describing diffusion processes [86–88]. Pervaporation with ceramic membranes is less well understood in terms of transport mechanisms. Consequently, modeling of ceramic pervaporation is still less mature, although the performance of the process was reported to be good [89]. Nomura et al. [90] studied the transport mechanism of ethanol/water through silicalite membranes in

3.4 Flux and Separation Prediction

pervaporation and vapor permeation and carried out single component and binary mixture adsorption experiments. It was shown that ethanol permeance was hardly inﬂuenced by the presence of water, whereas the water ﬂux decreased substantially in the presence of ethanol. An adsorption–diffusion model was considered for the transport through the membrane. The high-selective permeation of ethanol was explained by the ethanol-selective adsorption to the silicalite membrane. Krishna and Paschek [91] employed the Maxwell–Stefan description for mass transport of alkanes through silicalite membranes, but did not consider more complex (e.g., unsaturated or branched) hydrocarbons. Kapteijn et al. [92] and Bakker et al. [93] applied the Maxwell–Stefan model for hydrocarbon permeation through silicalite membranes. Flanders et al. [94] studied separation of C6 isomers by pervaporation through ZSM-5 membranes and found that separation was due to shape selectivity. 3.4.4 Hybrid Simulations

In industrial applications, pervaporation has to compete with conventional separation processes, such as distillation, liquid–liquid extraction, adsorption, and stripping. Pervaporation has attracted the interest of the chemical industry for separations that are difﬁcult to achieve by distillation, for example, separations giving azeotropic mixtures and separations of components with a small difference in volatility. Pervaporation as a standalone technique is still to be developed industrially, but as part of a hybrid process, combined with for example, distillation (Figure 3.3), it is very promising for difﬁcult separations and may yield considerable energy savings. Several authors have already developed methodologies for the simulation of hybrid distillation–pervaporation processes. Short-cut methods were developed by Moganti et al. [95] and Stephan et al. [96]. Due to simpliﬁcations such as the use of constant relative volatility, one-phase sidestreams, perfect mixing on feed and permeate sides of the membrane, and simple membrane transport models, the results obtained should only be considered qualitative in nature. Verhoef et al. [97] used a quantitative approach for simulation, based on simpliﬁed calculations in Aspen Plus/Excel VBA. H€ommerich and Rautenbach [98] describe the design and optimization of combined pervaporation–distillation processes, incorporating a user-written routine for pervaporation into the Aspen Plus simulation software. This is an improvement over most approaches with respect to accuracy, although the membrane model itself is still quite

Figure 3.3 Possible configurations for distillation–pervaporation hybrid processes.

j57

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

58

limited. Furthermore, most authors analyze and optimize the performance of only a particular, predetermined hybrid conﬁguration. From the literature references cited and deriving from one of the conclusions of Lipnizki et al. [85], it appears that more adequate and accurate process design tools to optimize hybrid techniques involving pervaporation are strongly needed.

3.5 Conclusions

Solvent-resistant nanoﬁltration and pervaporation are undoubtedly the membrane processes needed for a totally new approach in the chemical process industry, the pharmaceutical industry and similar industrial activities. This is generally referred to as process intensiﬁcation and should allow energy savings, safer production, improved cost efﬁciency, and allow new separations to be carried out. Problems to be solved are related to membrane stability (of polymeric membranes, but also the development of hydrophobic ceramic nanoﬁltration membranes and pervaporation membranes resistant to extreme conditions), to a lack of fundamental knowledge on transport mechanisms and models, and to the need for simulation tools to be able to predict the performance of solvent-resistant nanoﬁltration and pervaporation in a process environment. This will require an investment in basic and applied research, but will generate a breakthrough in important societal issues such as energy consumption, global warming and the development of a sustainable chemical industry.

Acknowledgment

The Research Council of the K.U. Leuven is gratefully acknowledged (OT/2006/37).

References 1 Seader, J.D. and Henley, E.J. (2006) Separation Process Principles, John Wiley & Sons, Hoboken, NJ. 2 Bernardo, P., Barbieri, G. and Drioli, E. (2006) Chemical Engineering Research & Design, 84 (A5), 405–411. 3 Macedonio, F., Curcio, E. and Drioli, E. (2007) Desalination, 203 (1–3), 260–276. 4 Mehdizadeh, H. (2006) Desalination, 191 (1–3), 200–209. 5 Reisch, M.S. (2007) Chemical & Engineering News, 85 (17), 22–24.

6 Rachwall, T. and Judd, S. (2006) Water Environ Journal, 20 (3), 110–113. 7 Van der Bruggen, B. and Vandecasteele, C. (2002) Desalination, 143 (3), 223–234. 8 Miller, G.W. (2006) Desalination, 187 (1–3), 65–75. 9 Wintgens, T., Melin, T., Sch€afer, A., Khan, S., Muston, M., Bixio, D. and Thoeye, C. (2005) Desalination, 178 (1–3), 1–11. 10 Into, M., Jonsson, A.S. and Lengden, G. (2004) Journal of Membrane Science, 242 (1–2), 21–25.

References 11 Karakulski, K., Gryta, M. and Sasim, M. (2006) Chemical Papers-Chemicke Zvesti, 60 (6), 416–421. 12 Kremser, U., Drescher, G., Otto, S. and Recknagel, V. (2006) Desalination, 189 (1–3), 53–58. 13 Pribic, P., Roza, M. and Zuber, L. (2006) Separation Science and Technology, 41 (11), 2581–2602. 14 White, L.S. (2002) Journal of Membrane Science, 205, 191–202. 15 Van der Bruggen, B., Geens, J. and Vandecasteele, C. (2002) Separation Science and Technology, 37 (4), 783–797. 16 Toh, Y.H.S., Loh, X.X., Li, K., Bismarck, A. and Livingston, A.G. (2007) Journal of Membrane Science, 291 (1–2), 120–125. 17 Tarleton, E.S., Robinson, C.P., Millington, C.R. and Nijmeijer, A. (2005) Journal of Membrane Science, 252 (1–2), 123–131. 18 Geens, J., De Witte, B. and Van der Bruggen, B. (2007) Separation Science and Technology, 42 (11), 2435–2449. 19 Aerts, S., Buekenhoudt, A., Weyten, H., Gevers, L.E.M., Vankelecom, I.F.J. and Jacobs, P.A. (2006) Journal of Membrane Science, 280 (1–2), 245–252. 20 Witte, P.T., Chowdhury, S.R., ten Elshof, J.E., Sloboda-Rozner, D., Neumann, R. and Alsters, P.L. (2005) Chemical Communications, 9, 1206–1208. 21 White, L.S. (2006) Journal of Membrane Science, 286 (1–2), 26–35. 22 Mulder, M. (1996) Basic Principles of Membrane Technology, 2nd edn, Kluwer Academic Publishers. 23 Joncquieres, A., Clement, R., Lochon, P., Neel, J., Dresch, M. and Chretien, B. (2002) Journal of Membrane Science, 206, 87–117. 24 Pereira, C.C., Ribeiro, C.C., Nobrega, R. and Borges, C.P. (2006) Journal of Membrane Science, 274 (1–2), 1–23. 25 Semenova, S.I., Ohya, H. and Soontarapa, K. (1997) Desalination, 110 (1–3), 251–286. 26 Vandezande, P., Gevers, L.E.M., Paul, J.S., Vankelecom, I.F.J. and Jacobs, P.A. (2005) Journal of Membrane Science, 250 (1–2), 305–310.

27 Cuperus, F.P. and Ebert, K. (2005) Nonaqueous applications of NF, Chapter 21 in Nanoﬁltration, Principles and Applications (eds A.I. Sch€afer, A.G. Fane and T.D. Waite), Elsevier, Oxford, New York and Tokyo. 28 Linder, C., Perry, M. and Ketraro, R. (1988) US Patent 4, 761, 233. 29 Golemme, G. and Drioli, E. (1996) Journal of Inorganic and Organometallic Polymers, 6 (4), 341–365. 30 Lundgard, R.A. (1996) US Patent 5, 507, 984. 31 Shimoda, T. and Hachiya, H. (1999) US Patent 5, 997, 741. 32 Van Gemert, R.W., Bos, A.A.C.M., Maass, L. and Bakker, W.J.W. (2001) Proceedings of the conference Engineering with Membranes, June 2001, Granada Spain. 33 Gevers, L.E.M., Aldea, S., Vankelecom, I.F.J. and Jacobs, P.A. (2006) Journal of Membrane Science, 281 (1–2), 741–746. 34 Gevers, L.E.M., Vankelecom, I.F.J. and Jacobs, P.A. (2006) Journal of Membrane Science, 278 (1–2), 199–204. 35 Bakker, W.J.W., Bos, A.C.M., Rutten, W.L.P., Keurentjes, J.T.F. and Wessling, M. (1998) Proceedings of the 5th International Conference on Inorganic Membranes, Nagoya June 22–26, pp. 448–451. 36 Morigami, Y., Kondo, M., Abe, J., Kita, H. and Okamoto, K. (2000) Book of Abstracts of the 6th International Conference on Inorganic Membranes, Montpellier, June 26–30, pp. 55. 37 Verkerk, A.W., Van Male, P., Vorstman, M.A.G. and Keurentjes, J.T.F. (2001) Journal of Membrane Science, 1–2 (193), 227–238. 38 Tsuru, T., Miyawaki, M., Kondo, H., Yoshioka, T. and Asaeda, M. (2003) Separation and Puriﬁcation Technology, 32 (1–3), 105–109. 39 Voigt, I., Fischer, G., Puhlfurss, P., Schleifenheimer, M. and Stahn, M. (2003) Separation and Puriﬁcation Technology, 32 (1–3), 87–91. 40 Van Gestel, T., Vandecasteele, C., Buekenhoudt, A., Dotremont, C.,

j59

j 3 Fundamentals of Membrane Solvent Separation and Pervaporation

60

41

42

43

44

45

46 47

48

49 50

51

52

53

Luyten, J., Leysen, R., Van der Bruggen, B. and Maes, G. (2002) Journal of Membrane Science, 207, 73–89. Voigt, I., Fischer, G., Puhlf€ urss, P. and Seifert, D. (1998) Proceedings of the 5th International Conference on Inorganic Membranes, Nagoya, Japan June 22–26, pp. 42–45. Burggraaf, A.J. and Keizer, K. (1991) Inorganic Membranes: Synthesis, Characterization and Applications (ed. R. Bhave), Van Nostrand Reinhold, New York. Van Gestel, T., Van der Bruggen, B., Buekenhoudt, A., Dotremont, C., Luyten, J., Vandecasteele, C. and Maes, G. (2003) Journal of Membrane Science, 224 (1–2), 3–10. Leger, C., De, H., Lira, L. and Paterson, R. (1996) Journal of Membrane Science, 120 (1–2), 187–195. Javaid, A., Hughey, M.P., Varutbangkul, V. and Ford, D.M. (2001) Journal of Membrane Science, 187 (1–2), 141–150. Miller, J.R. and Koros,W.J.(1990) Separation Science and Technology, 25, 1257–1280. Simon, C., Bredesen, R. and Denonville, C. (1998) Proceedings of the 5th International Conference on Inorganic Membranes (ICIM), Nagoya, Japan June 22–26, pp. 416–419. Picard, C., Larbot, A., Guida-Pietrasanta, F., Boutevin, B. and Ratsimihety, A. (2001) Separation and Puriﬁcation Technology, 25, 65–69. Randon, J. and Paterson, R. (1997) Journal of Membrane Science, 134 (1–2), 219–223. Caro, J., Noack, M. and K€olsch, P. (1998) Microporous and Mesoporous Materials, 22, 321–332. Picard, C., Larbot, A., Tronel-Peyroz, E. and Berjoan, R. (2004) Solid State Sciences, 6 (6), 605–612. Impens, N.R.E.N., Van der Voort, P. and Vansant, E.F. (1999) Microporous and Mesoporous Materials, 28 (2), 217–232. Hair, M.L. and Tripp, C.P. (1995) Colloids and Surfaces A: Physicochemical and Engineering, Aspects, 105 (1), 95–103.

54 Raman, L.P., Cheryan, M. and Rajagopalan, N. (1996) Fett-Lipid, 98 (1), 10–14. 55 Bridge, M.J., Broadhead, K.W., Hlady, V. and Tresco, P.A. (2002) Journal of Membrane Science, 195 (1–2), 51–64. 56 Subramanian, R., Nakajima, M. and Kawakatsu, T. (1998) Journal of Food Engineering, 38 (1), 41–56. 57 Van Gestel, T., Vandecasteele, C., Buekenhoudt, A., Dotremont, C., Luyten, J., Van der Bruggen, B. and Maes, G. (2003) Journal of Membrane Science, 214 (1), 21–29. 58 Ebert, K. and Cuperus, F.P. (1999) Membrane Technology, 107, 5–8. 59 Van der Bruggen, B., Jansen, J.C., Figoli, A., Geens, J., Van Baelen, D., Drioli, E. and Vandecasteele, C. (2004) The Journal of Physical Chemistry. B, 108, 13273–13279. 60 Khayet, M. and Matsuura, T. (2004) AIChE Journal, 50 (8), 1697–1712. 61 Sekulic, J., ten Elshof, J.E. and Blank, D.H.A. (2004) Advanced Materials, 16 (17), 1546. 62 Machado, D.R., Hasson, D. and Semiat, R. (2000) Journal of Membrane Science, 166, 63–69. 63 Yang, X.J., Livingston, A.G. and Freitas dos Santos, L. (2001) Journal of Membrane Science, 190, 45–55. 64 Silva, P., Han, S.J. and Livingston, A.G. (2005) Journal of Membrane Science, 262 (1–2), 49–59. 65 Paul, D.R. (2004) Journal of Membrane Science, 241, 371–386. 66 Bhanushali, D., Kloos, S., Kurth, C. and Bhattacharyya, D. (2001) Journal of Membrane Science, 189, 1–21. 67 Scarpello, J.T., Nair, D., dos Santos, L.M.F., White, L.S. and Livingston, A.G. (2002) Journal of Membrane Science, 203 (1–2), 71–85. 68 Gibbins, E., DAntonio, M., Nair, D., White, L.S., Freitas dos Santos, L.M., Vankelecom, I.F.J. and Livingston, A.G. (2002) Desalination, 147, 307–313. 69 Geens, J., Van der Bruggen, B. and Vandecasteele, C. (2006) Separation and

References

70

71

72

73

74

75 76

77 78

79

80

81

82

83

Puriﬁcation Technology, 48 (3), 255–263, Dordrecht, The Netherlands. Gevers, L.E.M., Meyen, G., De Smet, K., De Velde, P.V., Du Prez, F., Vankelecom, I.F.J. and Jacobs, P.A. (2006) Journal of Membrane Science, 274 (1–2), 173–182. Vankelecom, I.F.J., De Smet, K., Gevers, L.E.M., Livingston, A., Nair, D., Aerts, S., Kuypers, S. and Jacobs, P.A. (2004) Journal of Membrane Science, 231 (1–2), 99–108. Matsuura, T. and Sourirajan, S. (1981) Industrial & Engineering Chemistry Process Design and, Development, 20, 272–282. Santos, J.L.C., Hidalgo, A.M., Oliveira, R., Velizarov, S. and Crespo, J.G. (2007) Journal of Membrane Science, 300 (1–2), 191–204. Dijkstra, M.F.J., Bach, S. and Ebert, K. (2006) Journal of Membrane Science, 286 (1–2), 60–68. Zeman, L. and Wales, M. (1981) Separation Science and Technology, 16 (3), 275–290. Li, W., Li, J., Chen, T., Zhao, Z. and Chen, C. (2005) Journal of Membrane Science, 258, 8–15. Nakao, S.I. and Kimura, S. (1982) Journal of ChemicalEngineeringofJapan,15(3),200–205. Van der Bruggen, B., Schaep, J., Vandecasteele, C. and Wilms, D. (2000) Separation Science and Technology, 35 (2), 169–182. Tarleton, E.S., Robinson, J.P., Millington, C.R., Nijmeijer, A. and Taylor, M.L. (2006) Journal of Membrane Science, 278 (1–2), 318–327. Geens, J., Hillen, A., Bettens, B., Van der Bruggen, B. and Vandecasteele, C. (2005) Journal of Chemical Technology and Biotechnology (Oxford, UK, 1986), 80 (12), 1371–1377. Geens, J., Boussu, K., Vandecasteele, C. and Van der Bruggen, B. (2006) Journal of Membrane Science, 281 (1–2), 139–148. Tarleton, E.S., Robinson, J.P., Smith, S.J. and Na, J.J.W. (2005) Journal of Membrane Science, 261 (1–2), 129–135. Tarleton, E.S., Robinson, J.P. and Salman, M. (2006) Journal of Membrane Science, 280 (1–2), 442–451.

84 Lee, C.H. (1975) Journal of Applied Polymer Science, 19 (1), 83–95. 85 Lipnizki,F.,Hausmanns,S.,Ten,P.K.,Field, R.W. and Laufenberg, G. (1999) Chemical Engineering Journal, 73 (2), 113–129. 86 Lipnizki, F. and Tragardh, G. (2001) Separation and Puriﬁcation Methods, 30 (1), 49–125. 87 Wang, H.Y., Ugomori, T., Tanaka, K., Kita, H., Okamoto, K. and Suma, Y. (2000) Journal of Polymer Science Part B-Polymer Physics, 38 (22), 2954–2964. 88 Jonquieres, A., Perrin, L., Arnold, S., Clement, R. and Lochon, P. (2000) Journal of Membrane Science, 174 (2), 255–275. 89 Van Veen, H.M., van Delft, Y.C., Engelen, C.W.R. and Pex, P.P.A.C. (2001) Separation and Puriﬁcation Technology, 223 (1–3), 361–366. 90 Nomura, M., Yamaguchi, T. and Nakao, S. (1998) Journal of Membrane Science, 144 (1–2), 161–171. 91 Krishna, R. and Paschek, D. (2000) Separation and Puriﬁcation Technology, 21 (1–2), 111–136. 92 Kapteijn, F., Bakker, W.J.W., Zheng, G.H., Poppe, J. and Moulijn, J.A. (1995) Chemical Engineering Journal, 57 (2), 145–153. 93 Bakker, W.J.W., Kapteijn, F., Poppe, J. and Moulijn, J.A. (1996) Journal of Membrane Science, 117 (1–2), 57–78. 94 Flanders, C.L., Tuan, V.A., Noble, R.D. and Falconer, J.L. (2000) Journal of Membrane Science, 176, 43–53. 95 Moganti, S., Noble, R.D. and Koval, C.A. (1994) Journal of Membrane Science, 93, 31–44. 96 Stephan, W., Noble, R.D. and Koval, C.A. (1995) Journal of Membrane Science, 99, 259–272. 97 Verhoef, B.A., Huybrechs, B., van Veen, H., Pex, P., Degreve, J. and Van der Bruggen, B. (2008) Simulation of a hybrid pervaporation-distillation process, Computers & Chemical Engineering, 32 (6), 1135–1146. 98 H€ommerich, U. and Rautenbach, R. (1998) Journal of Membrane Science, 146, 53–64.

j61

j63

4 Fundamentals of Membrane Gas Separation Tom M. Murphy, Grant T. Offord, and Don R. Paul

4.1 Introduction

Research and technology innovations in the 1960s and 1970s led to the signiﬁcant commercial practice of gas separations by membranes that exist today. These advances involved developing membrane structures that could produce high ﬂuxes and modules for packaging large amounts of membrane area per unit volume. The discovery of asymmetric membrane structures for reverse osmosis was a key step in this evolution [1, 2]; such structures were eventually created in hollow ﬁbers using solution spinning technology. Typical asymmetric membranes exhibit defects upon drying that limit their value for gas separations; however, this problem was eventually solved by the discovery that the defects could be effectively sealed by coating the membrane with a highly permeable polymer, such as silicone rubber [3, 4]. Composite membranes consisting of a thin separating layer coated onto a porous substrate or an intermediate layer have also been developed, which has expanded the types of materials that can be converted into high-ﬂux membranes. For commercial use, high-ﬂux ﬂat-sheet membranes are packaged into spiral wound modules, while hollow-ﬁber membranes are assembled into modules resembling shell and tube heat exchangers. Today, such membrane modules are sold commercially by a number of companies for separating nitrogen from air, recovery of hydrogen from process streams, natural-gas processing, dehydration of gas streams, recovery of vapors from gases, and so on. A number of recent books summarize these developments and their industrial uses [5–8]. Most of the more recent research has focused on developing membrane materials with a better balance of selectivity and productivity (permeability) as that seems the most likely route for expanding the use of this technology. There appear to be natural upper bounds [9, 10] on this tradeoff that limit the extent of improvement that can be realized by manipulating the molecular structure of the polymer used for the selective layer of high-ﬂux membranes, at least in many cases. This has led to interest in nonpolymeric and so-called mixed-matrix materials for membrane formation [8]; however, at this time, polymers remain the materials of choice for gas-separation

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 4 Fundamentals of Membrane Gas Separation

64

membranes. The purposes of this chapter are to review brieﬂy the fundamentals of gas permeation in polymeric materials and to explore in some detail two very different material issues of current interest. One of these relates to the physical aging of glassy polymer membranes that results from their nonequilibrium character. The other relates to the search for membrane materials that have exceptional selectivity for CO2 relative to other gases, and it turns out that some of the most promising polymers are in the rubbery state.

4.2 Polymer Structure and Permeation Behavior

Most polymers that have been of interest as membrane materials for gas or vapor separations are amorphous and have a single phase structure. Such polymers are converted into membranes that have a very thin dense layer or skin since pores or defects severely compromise selectivity. Permeation through this dense layer, which ideally is defect free, occurs by a solution–diffusion mechanism, which can lead to useful levels of selectivity. Each component in the gas or vapor feed dissolves in the membrane polymer at its upstream surface, much like gases dissolve in liquids, then diffuse through the polymer layer along a concentration gradient to the opposite surface where they evaporate into the downstream gas phase. In ideal cases, the sorption and diffusion process of one gas component does not alter that of another component, that is, the species permeate independently. For rubbery polymers, that is, above the glass-transition temperature, Tg, the sorption of simple gases follows the relationship known as Henrys law C ¼ Sp

ð4:1Þ

where C is the equilibrium concentration of the gas dissolved in the polymer when its partial pressure in the gas phase is p and S is the solubility coefﬁcient. At steady state, the diffusion process is described by a simple version of Ficks law Flux ¼

DDC DSDp PDp ¼ ¼ l l l

ð4:2Þ

where l is the dense layer thickness, DC is the concentration difference of gas in the upstream and downstream faces of the dense layer and D is the diffusion coefﬁcient. Since by Equation 4.1, DC ¼ SDp, we can see that the permeability coefﬁcient P is given by P ¼ DS ð4:3Þ Polymers above their Tg are in a state of equilibrium much like simple liquids. However, upon cooling below Tg, polymers are not able to achieve an equilibrium state since the polymer chain segments lack sufﬁcient mobility to reach this state in realizable time scales. Thus, glassy polymers exist in a nonequilibrium state that is a function of the prior history of the sample. It is useful to think of simple volumetric thermal expansion where at equilibrium the speciﬁc volume at a given temperature and pressure is Veq(T, p); the speciﬁc volume of a rubbery polymer is given by Veq. The

4.2 Polymer Structure and Permeation Behavior

observed speciﬁc volume of a glassy polymer, Vg, will always be larger than Veq; the excess volume of the glass (Vg Veq) affects many of the characteristics of the material and depends on the prior history of the sample. Because of the difference in segmental mobilities, glassy polymers are more than 103 times stiffer than rubbery polymers. The sorption of simple gases in glassy polymers follows a more complex relation and is well described by the so-called dual sorption model [11–17] C ¼ kD p þ

C0 H bp 1 þ bp

ð4:4Þ

where kD, C 0 H and b are parameters of the model. The second term on the right in Equation 4.4 represents an additional mode of sorption that can be linked quantitatively to the excess volume of the glassy state (Vg Veq) while the ﬁrst term may be thought to represent an extension of the Henrys law mode seen above Tg [13, 14]. Thus, the extent of sorption of gases in glassy polymers is actually signiﬁcantly greater than in rubbery polymers, which is counterintuitive considering that glasses are orders of magnitude more stiff than rubbers. In addition, the amount of sorption in the glass depends on the history of the sample; both effects being attributable to the nonequilibrium character [15–17]. The permeation of simple gases in glassy polymers is more complex than in rubbery polymers. An extension of the dual sorption model of permeation leads to a relation, when the downstream pressure is small, of the following form P ¼ kD DD þ

C0 H bDH 1 þ bp2

ð4:5Þ

where DD and DH are diffusion coefﬁcients for gas molecules sorbed by each of the modes of sorption and p2 is the upstream gas pressure [11–17]. This model predicts that the permeability coefﬁcient decreases slightly as the upstream pressure is increased and generally describes experimental data quite well. Later, we will return to the issue of how history affects permeation behavior of glassy polymers of the type used to make gas-separation membranes like those shown in Table 4.1. For gases or vapors that are quite soluble in polymers some of the simple relations described above break down. For example, the sorption isotherm for vapors in rubbery polymers may show upward curvature from the simple linear prediction of Henrys law, Equation 4.1, and this effect is actually expected from thermodynamic theories like the well-known Flory–Huggins equation [8]. In addition, the presence of the penetrant at high enough concentrations will affect the mobility of the polymer segments, which will be reﬂected as an increase in the penetrant diffusion coefﬁcient; this is referred to as plasticization [8]. Similar effects can also be seen in glassy polymers; but as might be expected, they are even more complex to describe. In addition to plasticization, sorbing sizable quantities of penetrant into a glassy polymer alters the state of the glass such that after removal of the penetrant the glass does not return to its original state; this has been referred to as conditioning [17]. For now it is useful to take a more global view and not concern ourselves with the nonlinear effects described by Equations 4.4 and 4.5 or those caused by plasticization.

j65

j 4 Fundamentals of Membrane Gas Separation

66

Table 4.1 Three glassy polymers used to form gas-separation membranes.

Polysulfone (PSF)

Matrimid

Poly(2,6-dimethyl1,4-phenylene oxide) (PPO)

Thus, we can use Equation 4.3 as representative of the solution–diffusion mechanism where S and D may not be constants but depend on the external conditions in the gas phases. With this in mind for a pair of gases A and B, we can construct the following useful relationship PA DA SA ð4:6Þ ¼ PB DB SB The ratio PA/PB is often referred to as the permselectivity of the membrane, and in simple cases allows one to determine the extent of separation that a given membrane can achieve in a given situation. It is a property of the membrane material and does not depend on the thickness of the separating layer. This permselectivity is the product of the diffusion selectivity and the solubility selectivity. To understand these terms, it is useful to know that in the simplest of cases, the diffusion coefﬁcient of a penetrant in a given polymer decreases as the size of the penetrant molecule increases and that the solubility coefﬁcient increases as the condensability of the penetrant increases. This is illustrated schematically in Figure 4.1. Various molecular diameters, the van der Waals volume and the critical volume have been used to characterize penetrant size [7, 8]. The propensity of the penetrant to condense, that is, its condensability, may be characterized by its boiling point, critical temperature, or the Lennard-Jones potential well depth, e/k. It is now well established that the dependence of D on penetrant size is much stronger for glassy polymers than for rubbery polymers, as suggested in Figure 4.1; that is, glasses may be said to be more size selective. Thus, glassy polymers like those shown in Table 4.1 have become the materials of choice for membranes to separate certain gas pairs. However, as we will explore more fully later, there are cases where rubbery polymers are more selective. In many cases, but not all, the condensability of penetrants increases as size increases. This is the case for the gas pair i and j suggested in Figure 4.1. In the case shown there, the diffusion selectivity favors i over j but the solubility selectivity favors j over i.

4.2 Polymer Structure and Permeation Behavior

Figure 4.1 Schematic illustration of how the (a) diffusion coefficient of penetrants depend on their size in rubbery and glassy polymers and (b) solubility coefficients for penetrants depend on their condensability.

Table 4.2 illustrates the various selectivity factors for some typical rubbery polymers, that is, silicone rubber, poly(dimethyl siloxane), and natural rubber, polyisoprene, and a glassy polymer, polysulfone. Here, we consider the important O2/N2 pair and several pairs involving CO2 that will be our focus later. In all the cases, the solubility selectivity is greater than unity and there is not a large difference between rubbery and glassy polymers. For most of these pairs, the diffusion selectivity is greater than unity, but there are some exceptions for CO2/O2 and CO2/N2 that reﬂect Table 4.2 Permselectivity characteristics of selected rubbery and glassy polymers.

SA/SB

DA/DB

T C

Reference

Poly(dimethyl siloxane) O2/N2 2.0 CO2/O2 4.9 7.4 CO2/N2 CO2/CH4 3.1

1.6 3.4 8.1 2.9

1.3 1.4 0.91 1.1

35,25 35,20 35 35

[18, 19] [18, 19] [18, 20] [18]

Natural rubber O2/N2 CO2/O2 CO2/N2 CO2/CH4

2.9 5.6 16 4.5

2.0 8.0 16 3.6

1.4 1.4 1.0 1.2

25 25 25 25

[19] [19] [19] [19]

Polysulfone O2/N2 CO2/O2 CO2/N2 CO2/CH4

5.6 4.0 22 22

1.6 8.8 14 3.7

3.5 0.45 1.6 5.9

35 35 35 35

[6] [6] [6] [6]

A/B

PA/PB

j67

j 4 Fundamentals of Membrane Gas Separation

68

subtle shape issues that become important when the sizes are similar and will not be pursued here. The important point is to see the much greater diffusion selectivity for O2/N2 and CO2/CH4 in polysulfone than in the two rubbery polymers; this translates into greater permselectivity of the glassy material than of the rubbery ones. When the gas or vapor feed stream contains a component that is highly soluble in the polymer membrane and causes plasticization, then the selectivity as deﬁned by Equation 4.6 will depend on the partial pressure or the amount of the plasticizing component sorbed into the membrane. Furthermore, pure-gas permeation measurements are generally not a good indicator of the separation performance, and mixed-gas permeation measurements will be needed [21–23]. Often, the mixed-gas selectivity is less than predicted from pure-gas measurements [8]; however, the opposite has been observed [24]. Competitive sorption effects can also compromise the prediction of mixed-gas behavior from pure-gas measurements [25]. For gas pairs where each component is less condensable than CO2, like O2/N2, it is generally safe to conclude that the selectivity characteristics can be accurately judged from pure-gas permeabilities at all reasonable pressures. When the gas pair involves a component more condensable than CO2, plasticization is likely to be a factor and pure-gas data may not adequately reﬂect mixed-gas selectivity. When CO2 is a component, the situation depends on the partial pressures and the nature of the polymer. Generally, polymers that crystallize are not considered good candidates for membrane materials; however, there are some exceptions [26, 27]. The presence of crystallinity reduces permeability [28, 29] and good membranes should be capable of high ﬂuxes. The usual physical picture is to think of a semicrystalline polymer in terms of a simple two-phase model; one phase being amorphous and the other being crystalline. In the typical case, the crystals do not sorb or transmit penetrant molecules; the following relationship has been proposed [28, 29] to describe the extent to which crystallinity reduces permeability from that if the polymer were amorphous Pc ¼

Pa ð1fÞ tb

ð4:7Þ

where Pc is the penetrant permeability in the semicrystalline polymer, Pa is the permeability of the completely amorphous polymer, f is the volume fraction of crystals, t is the tortuosity factor to account for a more elongated path a penetrant molecule must take through the amorphous phase since it cannot go through the crystals, and b is the chain-immobilization factor that reﬂects the reduction in mobility that occurs in the amorphous phase caused by the presence of crystallites. Ideally, neither f nor t will depend on what the penetrant is; however, b clearly can [29, 30]. Thus, it is possible that in addition to reducing permeability, crystallinity can alter selectivity, that is, bB ð4:8Þ ðPC ÞA =ðPC ÞB ¼ ðPa ÞA =ðPa ÞB bA Polymer blends and block-copolymers have been considered as membrane materials as mentioned later. If the components are miscible and a single-phase

4.3 Membranes from Glassy Polymers: Physical Aging

material results, then no special considerations are needed for applying the concepts outlined above. However, these systems usually consist of separate phases of the components. In this case, the materials need to be treated as a composite and, then, their morphology becomes an important issue [31, 32]. Usually, the continuous phase dominates the permeation process, so if we want to build into a membrane material the permeation characteristics of one component, then this component must have some degree of phase continuity in the material [33].

4.3 Membranes from Glassy Polymers: Physical Aging

Glassy polymers are usually the preferred materials for practical gas-separation membranes because of their inherently better permeability/selectivity balance than is typically the case for polymers above their glass-transition temperature [5, 9]. In addition, the structural rigidity provided by the glassy state is essential for membranes that must be self-supporting (e.g., asymmetric hollow ﬁbers) [5]. Glasses are not in a state of equilibrium; therefore, their properties are dependent on the details of their fabrication and time–temperature history [34–36]. Thus, it is not surprising to observe some variance in the reported properties, such as density, refractive index, gas permeability, and so on, of glassy polymers. At least for macroscopic specimens, the variability seems to be within a range small enough that meaningful property tabulations can be made for glassy polymers, as recorded in many handbooks [37]. However, recent research has shown that this variability may be considerably more pronounced for thin ﬁlms because of their signiﬁcantly more rapid evolution toward the equilibrium state, a process known as physical aging, most often observed in terms of volume relaxation or densiﬁcation [38]. This densiﬁcation, or physical aging, affects properties that are sensitive to free volume, such as permeability, and the associated changes can be quite signiﬁcant, on time scales of weeks to years [38–52]. Practical membranes must be very thin to achieve the high ﬂuxes needed for economical productivity; typically, skins or separating layers with thicknesses of the order of 0.1 mm (or 100 nm) or less with minimal defects are essential for a viable technology. However, such thin layers of glassy polymers can be greatly affected by the physical aging issues mentioned previously. This brings into question the widely practiced approach of using relatively thick ﬁlms for screening or selecting polymers as membrane materials. Indeed, the permeation properties of thick ﬁlms are often used to calculate the effective thickness of the skin layer of asymmetric or composite membrane structures from observed ﬂuxes. Figure 4.2 is an attempt to classify glassy polymer ﬁlms into different regimes of behavior according to thickness. To the far right of the thickness scale is the familiar case where properties, including those related to the departure from an equilibrium state, are expected to be independent of specimen size. This is clearly the expectation on the millimeter or centimeter scale and probably extends down to several micrometers; we might call this the bulk regime. On the other extreme are ultrathin ﬁlms where the thickness is of the same order of magnitude as the dimensions of the

j69

j 4 Fundamentals of Membrane Gas Separation

70

Figure 4.2 Approximate thickness regimes for the behavior of glassy polymer films [52]. Reproduced with permission of the American Chemical Society.

polymer chain coils (<100 nm, typically). In this region, the conformations of the polymer chains are perturbed by the boundaries imposed by the surfaces of the ﬁlms, which results in the so-called conﬁnement effects. The recent literature on polymer physics contains many experimental and theoretical studies of such ultrathin ﬁlms [53–56]. The glass-transition temperature (Tg) and other characteristics of ultrathin ﬁlms have been reported to be dependent on thickness; certain differences in behavior have been attributed to whether the ﬁlm is freestanding or supported on some substrate, which may inﬂuence the polymer. It is beyond the scope of this chapter to review or explain this regime. It is sufﬁcient for current purposes to note that characteristics such as Tg seem to reach a plateau or bulk value prior to thicknesses of the order of 100 nm, which is on the order of the thickness of the separating skin layer of some commercial asymmetric membranes. There is a region of thin ﬁlms with thicknesses between the two previously described extreme limits, ranging from 100 nm to several micrometers, where volume relaxation processes – and, hence, the change in gas-permeability properties with time – are much more rapid than that expected based on observations of bulk specimens as shown below. The results shown next are for thin ﬁlms prepared over a range of thicknesses from the polymers shown in Table 4.1 by spin casting that were then heated above Tg brieﬂy to erase prior history and then cooled to 35 C where aging was observed for more than a year [45]. Ellipsometry was used to measure the ﬁlm thickness and refractive index accurately [38, 45]. Figure 4.3 shows how the permeability of O2 in PPO decreases with aging time at 35 C for ﬁlms that range in thickness from 400 nm to 1 mil. The dashed line shows the single bulk value for O2 permeability reported in the literature for PPO measured on thick ﬁlms. At short aging times, the permeability values reported lie well above the so-called bulk value. This is believed to reﬂect the state of relatively high free volume captured by the protocol of ﬁlm preparation used in this work. At longer aging times, the current permeability values decrease well

4.3 Membranes from Glassy Polymers: Physical Aging

Figure 4.3 Oxygen permeability coefficients of poly (2,6-dimethyl1,4-phenylene oxide) (PPO) films of various thicknesses, as a function of aging time at 35 C [52]. Reproduced with permission of the American Chemical Society.

below the bulk values except for some thick ﬁlms. Similar results have been observed for the other polymers shown in Table 4.1 as well as several novel polyimides [49, 50]. Thus, the central conclusion is that the bulk values may differ from the timedependent values for thin ﬁlms by very signiﬁcant amounts. This results from the well-known fact that the state of a glassy polymer is dependent on its prior history, and the lesser-known fact that the rate of physical aging of thin ﬁlms can be quite signiﬁcant at temperatures well below the Tg value; in the present case, PSF is 150 C, PPO is 175 C, and Matrimid is 275 C below their respective Tg values. The aging rate is clearly dependent on ﬁlm thickness as shown in Figure 4.4. The thinnest ﬁlms examined here are of the order of 0.4 mm (or 400 nm) in thickness. One would expect even faster rates at 100 nm; however, current techniques have not yet permitted probing such thin samples because of issues of manipulating such fragile structures. The present results are not due to any alteration of the Tg value by thickness because such effects seem to be signiﬁcant only well below 100 nm [45, 53, 54]. Thus, it seems that different issues are at play here than in the regime we labeled ultrathin ﬁlms in Figure 4.2. The rate of change in permeability during aging is dependent on the size of the gas molecule; the rate of change for all the polymers studied follows the order O2 < N2 CH4 [48]. This is understandable because the underlying issue is the loss of free volume during aging and this has a larger effect on the permeation of larger molecules. Thus, the selectivity should increase on aging that is demonstrated to be the case in Figure 4.5 for PSF; the rate of increase is greater the thinner the ﬁlm. Clearly, the use of bulk values of gas permeability provides only a ﬁrst-order

j71

j 4 Fundamentals of Membrane Gas Separation

72

Figure 4.4 Effect of thickness on aging rate of glassy polymer films determined by change in oxygen permeability at 35 C [46]. Reproduced with permission of Elsevier.

Figure 4.5 O2/N2 permselectivity of polysulfone films of various thicknesses versus aging time at 35 C [46]. Reproduced with permission of Elsevier.

4.3 Membranes from Glassy Polymers: Physical Aging

approximation to the productivity/selectivity balance observed in thin ﬁlms because of the aging phenomenon. In addition to thickness, ellipsometry techniques also give the refractive index of thin ﬁlms [38, 45], which provides another useful way of tracking aging since refractive index can be related to density via relations like the Lorentz–Lorenz equation [38]. Figure 4.6 illustrates the change in refractive index (normalized by the initial value for thin ﬁlms (400 nm) of the three glassy polymers from Table 4.1. The increase in refractive index conﬁrms that the aging process involves densiﬁcation of the glass polymer. It is clear from these results that the aging rate of these three polymers is PPO > Matrimid>PSF, which is consistent with the results in Figure 4.4 where the aging response was tracked in terms of oxygen permeation. A more formal way to make the comparison between aging responses by permeation and optical properties is to deﬁne an aging rate as follows 1 qV qlnr ¼ ð4:9Þ r¼ V qlnt qlnt where V ¼ speciﬁc volume and r ¼ density of the polymer. The refractive-index data can be used to compute this aging rate using the Lorentz-Lorenz equation [38, 57]. Figure 4.7 compares the oxygen, nitrogen, and methane permeability reduction rates vs. the corresponding volumetric relaxation rate for thin ﬁlms of the 6FDA-based polyimides aged at 35 C with other thin glassy polymer ﬁlms, viz., polysulfone (PSF), the commercial polyimide Matrimid, and poly (2,6-dimethyl-1,4-phenylene oxide) (PPO). Clearly, there is a strong correlation between the two measures of aging rate

Figure 4.6 Normalized refractive indices for thin films (400 nm) of three glassy polymers as a function of aging time [38]. Reproduced with permission of the American Chemical Society.

j73

j 4 Fundamentals of Membrane Gas Separation

74

Figure 4.7 Correlation between O2, N2 and CH4 permeability reduction rates and their volumetric relaxation rates for thin films of various glassy polymers [49]. Reproduced with permission of Elsevier.

that is consistent among all these polymers. Note that the multiple points for the latter three polymers from the literature correspond to ﬁlms of different thickness. Interestingly, the 6FDA-based polyimide thin ﬁlms show much larger aging rates than the other materials, which is consistent with the higher free volume of these polyimides. The experimental results described above show that the gas-permeability properties of thin glassy polymer ﬁlms (submicrometer in thickness) are more time- or history-dependent than much thicker ﬁlms (the bulk state; for example, 50 mm or thicker) seem to be. This is manifested in terms of physical aging over a period of 1 year and more. The observed permeability values for the current thin ﬁlms are all initially greater than the reported bulk values but approach or become less than these values after a few days or weeks, depending on the thickness. After a year, the thin ﬁlms may be as much as four times less permeable than the reported bulk values. Selectivity increases with aging time, as might be expected from a densiﬁcation process. These observations have several practical consequences for membrane processes where the selective layers are as thin as or even thinner than the low end of the range studied here. First, it is clear that use of thick ﬁlm data to design or select membrane materials only gives a rough approximation of the performance that might be realized in practice. Second, because the absolute permeability of a thin ﬁlm may be severalfold different than the bulk permeability, use of the latter type of data to estimate skin thickness from ﬂux observations on asymmetric or composite membranes structures is also a very approximate method. Finally, these data indicate that one could expect

4.4 Membranes from Rubbery Polymers: Enhanced CO2 Selectivity

the productivity of commercial membrane modules to decline measurably over their lifetime of several years because of physical aging effects. Although there seems to be no published data showing such effects for gas-separation membranes based on polymer glasses, this is certainly the anecdotal experience of membrane suppliers and users. There is at least one documented report of such declines in reverseosmosis membranes [58]. Often, these declines for asymmetric membranes have been attributed to compaction of the porous substructure over time due to stress, effectively making the skin thicker; however, it is quite likely that at least some portion of this can be attributed to the more fundamental issue of physical aging of the glassy polymer skin. It should be said that one cannot precisely compare data such as that shown here to membrane module performance because the thin layers probably have experienced very different histories. In the present case, the ﬁlms were heated above the glass-transition temperature (Tg) and then cooled to ambient conditions to give a well-deﬁned starting state for the aging process. The skins of practical membranes are formed by a more complex process, and, generally, the online ﬂux monitoring begins after considerable aging has already occurred.

4.4 Membranes from Rubbery Polymers: Enhanced CO2 Selectivity

There are many examples of gas streams containing CO2 as an impurity that must be removed from lighter gases like CH4, N2, and H2. Examples include natural gas, where CH4 is the desired product, reﬁnery and reforming streams, where H2 is the valued product, ﬂue gases, where CO2 needs to be removed from N2 and sequestered, and others [59]. In these cases, membranes are needed that are much more permeable to CO2 relative to these light gases than can be found in conventional polymers (see the data in Table 4.2). Membranes are also useful in modiﬁed atmosphere packaging of fruits and vegetables for extending shelf life, and there is a need in certain applications for membranes that are more permeable to CO2, relative to O2 and N2, than current membrane materials [60]. This application will be explained more fully later. One strategy for designing membranes that are more selective to CO2 is to take advantage of its potential quadrapole interaction with the polymer to increase its solubility selectivity relative to the light gases, which cannot interact in this manner [59]. However, the advantages gained by building in this form of CO2 selectivity would not be fully realized if it were counterbalanced with a large size selectivity favoring the light gas; that is, one can expect glassy polymers to favor the light gas because of their larger diffusive selectivity as illustrated in Figure 4.1. For rubbery polymers, the size selectivity is minimal, as shown in Table 4.2, which gives the solubility selectivity a chance to dominate. As it turns out, poly(ethylene oxide), PEO, segments have an excellent afﬁnity for CO2 relative to O2, N2, CH4, and so on [20, 24, 59, 61–79]; however, poly(ethylene oxide) itself is highly crystalline, which reduces its permeability to all gases, it has a low melting point, 65 C, and it is water soluble. Thus, PEO is not directly useful for most of the applications mentioned, but there has been

j75

j 4 Fundamentals of Membrane Gas Separation

76

Table 4.3 Infinite dilution permselectivity characteristics of

semicrystalline poly(ethylene oxide) at 35 C [76]. A/B

PA/PB

SA/SB

DA/DB

O2/N2 CO2/O2 CO2/N2 CO2/CH4 CO2/H2

2.7 18 48 20 6.8

1.14 9 10.3 4.7 —

2.4 1.9 4.7 4.3 —

considerable research over the years directed at taking advantage of its desirable attributes, while minimizing its undesirable features [20, 24, 59, 61–79]. The various approaches to the problem outlined above have included blending PEO with polymers with which it may be miscible [61, 66], making block-copolymers of PEO with oligomers or polymers where the other segments were polyimides [64, 65, 67, 68], polyurethanes [71], polyamides [62, 70, 72–74] and polyesters [75], and making highly crosslinked structures containing ethylene oxide units [24, 76, 78, 79]. In general, these approaches can suppress PEO crystallization, prevent solubility in water, increase strength, and some lead to structures that can be converted to highﬂux composite hollow-ﬁbers [68] and ﬂat-sheet [62] membranes. The block-copolymers were designed to have high PEO contents to give a PEO continuous phase. Selected results from these studies are presented here to show how effective this approach can be. Table 4.3 shows the permselectivity characteristics of pure, semicrystalline PEO ﬁlms [76]. The selectivity characteristics for O2/N2 are rather similar to those for silicone rubber and natural rubber shown in Table 4.2. However, the values of permselectivity for CO2 relative to the various light gases shown are all much higher than Table 4.2 shows for the rubbery polymers listed there and even for polysulfone except for CO2/CH4. Comparison of the data in Tables 4.2 and 4.3 makes it clear that this high permselectivity of PEO stems from its high solubility selectivity for CO2 versus other gases; this is augmented by modest values of diffusivity selectivity. Data in Table 4.4 for the CO2/N2 pair illustrate that this effect can be translated into various block-copolymer structures when the PEO content is high enough to ensure it is the continuous phase. In fact, nearly all these materials have higher permselectivity and solubility selectivity for CO2/N2 than does pure PEO (see Table 4.3); however, the diffusion selectivity for these copolymers is much closer to, or even less than, unity than seen for pure PEO. Furthermore, the copolymers all have much higher absolute permeability coefﬁcients than does PEO. Figure 4.8 shows graphically how the permselectivity for CO2/N2 is much higher for any other gas relative to N2 in a poly(butylene terephthalate) block-copolymer containing 56 wt.% PEO segments [75]. The gas-solubility data in a polyamide blockcopolymer containing 57 wt.% PEO given in Figure 4.9 clearly demonstrate the unusually high CO2 solubility in these materials relative to other gases [70]. In fact, the CO2 data fall higher than expected based on the trend line set by the other gases by a factor of about 6. Apparently this effect is unique to ethylene oxide segments since

4.4 Membranes from Rubbery Polymers: Enhanced CO2 Selectivity Table 4.4 Carbon dioxide (A)/nitrogen (B) permselectivity characterization for polyimide, polyurethane and polyamide block-copolymers containing polyether segments.

Polymer

[PE]a wt%

PA/PB

SA/SB

DA/DB

T C

Reference

BP-ODA/DABA/PEO BP-ODA/DABA/PPO BP-ODA/DABA/PTHF MDI-BPA/PEO N6/PEO N6/PEO N12/PEO

57 57 58 60 70 57 55

69 28 29 47 71 56 51

77 33 — 41 26 88 63

0.92 0.84 — 1.15 2.8 0.63 0.81

25 25 25 35 25 35 35

[65] [65] [65] [71] [73] [72] [72]

a Weight per cent polyether in copolymer. PEO ¼ poly(ethylene oxide). PPO ¼ poly(propylene oxide). PTHF ¼ poly(tetramethylene oxide). BP-ODA/DABA ¼ polyimide (see Ref. 65 for structure). MDI-BPA ¼ polyurethane (see Ref. 71 for structure). N6 ¼ polyamide 6. N12 ¼ polyamide 12.

the data in Table 4.4 suggest that propylene oxide or tetramethylene oxide segments do not lead to such high CO2/N2 permselectivity characteristics. Next, we show how membranes with very high CO2 permselectivity relative to O2 and N2 would have value in preserving the shelf life of fruits and vegetables. It is well known that controlling the CO2 and O2 atmosphere around produce combined with refrigeration can extend the post-harvest viable life of produce; see Table 4.5 for recommended atmospheres for selected items of produce. The inset in Figure 4.10 illustrates how membranes are used to create a modiﬁed atmosphere inside a package of respiring produce. The produce consumes O2 and gives off CO2 in a

Figure 4.8 Selectivity for various gases relative to oxygen at 20 C for a PBT-PEO block-copolymer [75].

j77

j 4 Fundamentals of Membrane Gas Separation

78

Figure 4.9 Gas solubility in a polyamide-PEO block-copolymer containing 57 wt.% PEO units [70].

certain molar ratio or respiratory quotient (RQ) that normally is about 1.2; a high-ﬂux membrane patch afﬁxed to the package regulates the inﬂow of O2 and outﬂow of CO2. A theory for this process shows that at steady state the content of O2 and CO2 will lie on a line like those illustrated in Figure 4.10 that depend only on the permselectivity characteristics of the membrane and RQ [60]. Figure 4.10 shows lines for membranes that are nonselective and ones that have the properties of silicone rubber and poly (ethylene oxide). Exactly where on these lines the atmosphere in a given package will lie depends on the ratio of respiration to permeation rates and is a design parameter controlled by factors like membrane area and permeance plus the type and amount of produce in the package. Clearly, a silicone-rubber membrane combined with some nonselective perforations will meet the optimum requirement for many items of produce in Table 4.5 [60]. However, some items, like apples and pears, require a lower content of CO2 in the steady-state atmosphere than can be generated by such membranes. In these cases, a more CO2-selective membrane is needed, and those based on PEO segments in a suitable form to meet other requirements appear promising. Table 4.5 Recommended atmospheres for prolonging viable life of selected fruits and vegetables.a

Produce

T C

% O2

% CO2

Broccoli Cabbage Celery Oranges Strawberries Carrots (sliced) Lettuce (Iceberg, chopped) Apples (Braeburn) Pears (Bartlett)

0–5 0–5 0–5 5–10 0–5 0–5 0–5 0.7 1–0

1–2 2–3 1–4 5–10 5–10 2–5 0.5–3 1.8 1–2

5–10 3–6 3–5 0–5 15–20 15–20 10–15 1.0 0–1.5

a

From http://postharvest.ucdavis.edu.

References

Figure 4.10 Steady-state relationships between O2 and CO2 content for a modified atmosphere package outfitted with different membranes. Lines calculated by theory [60].

4.5 Summary

The chemical structure and physical state of the polymer has a considerable effect on how a membrane formed from it performs for gas separations. The glassy state is preferred for high selectivity when there are considerable differences in penetrant sizes; however, thin, glassy polymer layers or skins may undergo substantial decline in permeability over time owing to physical aging that is much more rapid than observed in bulk. Rubbery polymers allow the opportunity to base selectivity on penetrant solubility, as illustrated for CO2 relative to other gases. Polymers containing poly (ethylene oxide) segments have attractive CO2 selectivities because of their interactions with the CO2 quadrapole. However, to achieve practical membranes from such materials, it is necessary to suppress PEO crystallization (to achieve high ﬂuxes) and to retard swelling by water using crosslinking, blending, or block-copolymer structures.

Acknowledgments

This work was supported by the National Science Foundation under Grant No. DMR0423914.

References 1 Merten, U. (ed.) (1966) Desalination by Reverse Osmosis, MIT Press, Cambridge. 2 Riley, R.L., Gardner, J.O. and Merten, U. (1964) Science, 143, 801–803.

3 Henis, J.M.S. and Tripodi, M.K. (1980) Separation Science and Technology, 15, 1059. 4 Henis, J.M.S. and Tripodi, M.K. (1980) U.S. Patent 4,230,236, Monsanto.

j79

j 4 Fundamentals of Membrane Gas Separation

80

5 Kesting, R.E. and Fritzsche, A.K. (1993) Polymeric Gas Separation Membranes, John Wiley & Sons, New York. 6 Paul, D.R. and Yampolskii, Y.P. (1994) Polymeric Gas Separation Membranes, CRC Press, Boca Raton. 7 Baker, R.W. (2004) Membrane Technology and Applications, 2nd edn, John Wiley, Chichester. 8 Yampolskii, Y., Pinnau, I. and Freeman, B.D. (eds) (2006) Materials Science of Membranes for Gas and Vapor Separation, John Wiley, Chichester. 9 Robeson, L.M. (1991) Journal of Membrane Science, 62, 165–185. 10 Freeman, B.D. (1999) Macromolecules, 32, 375–380. 11 Paul, D.R. and Koros, W.J. (1976) Journal of Polymer Science: Polymer Physics Edition, 14, 687. 12 Koros, W.J., Chan, A.H. and Paul, D.R. (1977) Journal of Membrane Science, 2, 165. 13 Koros, W.J. and Paul, D.R. (1978) Journal of Polymer Science: Polymer Physics Edition, 16, 1947 and 2171. 14 Koros, W.J. and Paul, D.R. (1981) Journal of Polymer Science: Polymer Physics Edition, 19, 1655. 15 Chan, A.H. and Paul, D.R. (1979) Journal of Applied Polymer Science, 24, 1539. 16 Chan, A.H. and Paul, D.R. (1980) Polymer Engineering and Science, 20, 87. 17 Wonders, A.G. and Paul, D.R. (1979) Journal of Membrane Science, 5, 63. 18 Stern, S.A., Shah, V.M. and Hardy, B.J. (1987) Journal of Polymer Science, Part B: Polymer Physics Edition, 25, 1263–1298. 19 van Amerongen, G.J. (1964) Rubber Chemistry and Technology, 37, 1065–1152. 20 Hirayama, Y., Kase, Y., Tanihara, N., Sumiyama, Y., Kusuki, Y. and Haraya, K. (1999) Journal of Membrane Science, 160, 87–99. 21 Barbari, T.A., Koros, W.J. and Paul, D.R. (1989) Journal of Membrane Science, 42, 69. 22 Puleo, A.C., Paul, D.R. and Kelley, S.S. (1989) Journal of Membrane Science, 47, 301.

23 Raymond, P.C., Koros, W.J. and Paul, D.R. (1993) Journal of Membrane Science, 77, 49. 24 Lin, H., Van Wagner, E., Freeman, B.D., Toy, L.G. and Gupta, R.P. (2006) Science, 311, 639–642. 25 Koros, W.J., Chern, R.T., Stannett, V. and Hopfenberg, H.B. (1981) Journal of Polymer Science: Polymer Physics Edition, 19, 1513–1530. 26 Puleo, A.C., Paul, D.R. and Wong, P.K. (1989) Polymer, 30, 1357. 27 Mohr, J.M. and Paul, D.R. (1991) Polymer, 32, 1236. 28 Michaels, A.S. and Bixler, H.J. (1961) Journal of Polymer Science, 50, 393–412. 29 Michaels, A.S. and Bixler, H.J. (1961) Journal of Polymer Science, 50, 413–439. 30 Mogri, Z. and Paul, D.R. (2001) Polymer, 42, 7765–7780 and 7781–7789. 31 Bucknall, C.B. and Paul, D.R. (2000) Polymer Blends: Formulation and Performance, vols. 1 and 2, John Wiley and Sons, New York. 32 Hadjichristidis, N., Pispas, S. and Floudas, G.A. (2003) Block Copolymers: Synthetic Strategies, Physical Properties, and Applications, John Wiley and Sons, Hoboken. 33 Newman, S. and Paul, D.R. (1978) Polymer Blends, vols. 1 and 2, Academic Press, New York. 34 Struik, L.C.E. (1978) Physical Aging in Amorphous Polymers and Other Materials, Elsevier Scientiﬁc Publishing Co, New York. 35 Hutchinson, J.M. (1997) Progress in Polymer Science, 125, 23. 36 Curro, J.G., Lagasse, R.R. and Simha, R. (1982) Macromolecules, 15, 1621. 37 van Krevelen, D.W. (1990) Thermophysical properties of polymers, in Properties of Polymers: Their Correlation with Chemical Structure, Their Numerical Estimation and Prediction from Additive Group Contributions, Elsevier, Amsterdam and New York. 38 Huang, Y. and Paul, D.R. (2006) Macromolecules, 39, 1554.

References 39 Rezac, M.E., Pfromm, P.H., Costello, L.M. and Koros, W.J. (1994) Industrial & Engineering Chemistry Research, 32, 1921. 40 Pfromm, P.H. and Koros, W.J. (1995) Polymer, 36, 2379–2387. 41 Dorkenoo, K.D. and Pfromm, P.H. (1999) Journal of Polymer Science Part B-Polymer Physics, 37, 2239–2251. 42 Chung, T.S. and Teoh, S.K. (1999) Journal of Membrane Science, 152, 175. 43 Dorkenoo, K.D. and Pfromm, P.H. (2000) Macromolecules, 33, 3747. 44 McCaig, M.S. and Paul, D.R. (2000) Polymer, 41, 629–637 and 639–648. 45 Huang, Y. and Paul, D.R. (2004) Journal of Membrane Science, 244, 167–178. 46 Huang, Y. and Paul, D.R. (2004) Polymer, 45, 8377–8393. 47 Huang, Y. and Paul, D.R. (2005) Macromolecules, 38, 10148–10154. 48 Huang, Y., Wang, X. and Paul, D.R. (2006) Journal of Membrane Science, 277, 219–229. 49 Kim, J.H., Koros, W.J. and Paul, D.R. (2006) Polymer, 47, 3094–3103 and 3104–3111. 50 Kim, J.H., Koros, W.J. and Paul, D.R. (2006) Journal of Membrane Science, 282, 21–31 and 32–43. 51 Huang, Y. and Paul, D.R. (2007) Journal of Polymer Science Part B-Polymer Physics, 45, 1390–1398. 52 Huang, Y. and Paul, D.R. (2007) Industrial & Engineering Chemistry Research, 46, 2342–2347. 53 Kim, J.H., Jang, J. and Zin, W.C. (2000) Langmuir, 16, 4064. 54 Pham, J.Q. and Green, P.F. (2003) Macromolecules, 36, 1665–1669. 55 Forrest, J.A., Dalnoki-Veress, K., Stevens, J.R. and Dutcher, J.R. (1996) Physical Review Letters, 77, 2002. 56 Park, C.H., Kim, J.H., Ree, M., Sohn, B.H., Jung, J.C. and Zin, W.C. (2004) Polymer, 45, 4507. 57 Rowe, B.W., Freeman, B.D. and Paul, D.R. (2007) Macromolecules, 40, 2806. 58 Baayens, L. and Rosen, S.L. (1972) Journal of Applied Polymer Science, 16, 663.

59 Lin, H. and Freeman, B.D. (2005) Journal of Molecular Structure, 739, 57–74. 60 Paul, D.R. and Clarke, R. (2002) Journal of Membrane Science, 208, 269–283. 61 Kawakami, M., Iwanaga, H., Hara, Y., Iwamoto, M. and Kagawa, S. (1982) Journal of Applied Polymer Science, 27, 2387–2393. 62 Blume, I. and Pinnau, I. (1990) U.S. Patent 4,963,165, Membrane Technology Research, Inc. 63 Qipeng, G., Hechang, X. and Dezhu, M. (1990) Journal of Applied Polymer Science, 39, 2321–2330. 64 Okamoto, K., Umeo, N., Okamyo, S., Tanaka, K. and Kita, H. (1993) Chemistry Letters, 22, 225–228. 65 Okamoto, K., Fujii, M., Okamyo, S., Suzuki, H., Tanaka, K. and Kita, H. (1995) Macromolecules, 28, 6950–6956. 66 Li, J., Nagri, K., Nakagawa, T. and Wang, S. (1995) Journal of Applied Polymer Science, 58, 1455–1463. 67 Okamoto, K., Yasugi, N., Kawabata, T., Tanaka, K. and Kita, H. (1996) Chemistry Letters, 25, 613–614. 68 Suzuki, H., Tanaka, K., Kita, H., Okamoto, K., Hoshino, H., Yoshinaga, T. and Kusuki, Y. (1998) Journal of Membrane Science, 146, 31–37. 69 Tsutsui, K., Yoshimizu, H., Tsujita, Y. and Kinoshita, T. (1999) Journal of Applied Polymer Science, 73, 2733–2738. 70 Bondar, V.I., Freeman, B.D. and Pinnau, I. (1999) Journal of Polymer Science Part BPolymer Physics, 37, 2463–2475. 71 Yoshino, M., Ito, K., Kita, H. and Okamoto, K. (2000) Journal of Polymer Science Part BPolymer Physics, 38, 1707–1715. 72 Bondar, V.I., Freeman, B.D. and Pinnau, I. (2000) Journal of Polymer Science Part BPolymer Physics, 38, 2051–2062. 73 Kim, J.H., Ha, S.Y. and Lee, Y.M. (2001) Journal of Membrane Science, 190, 179–193. 74 Barbi, V., Funari, S.S., Gehrka, R., Scharnagl, N. and Stribeck, N. (2003) Macromolecules, 36, 749–758. 75 Metz, S.J., Mulder, M.H.V. and Wessling, M. (2004) Macromolecules, 37, 4590–4597.

j81

j 4 Fundamentals of Membrane Gas Separation

82

76 Lin, H. and Freeman, B.D. (2004) Journal of Membrane Science, 239, 105–117. 77 Lin, H., Van Wagner, E., Swinnea, J.S., Freeman, B.D., Pas, S.J., Hill, A.J., Kalakknunnath, S. and Kalika, D.S. (2006) Journal of Membrane Science, 276, 145–161.

78 Lin, H. and Freeman, B.D. (2006) Macromolecules, 39, 3568–3580. 79 Lin, H., Van Wagner, E., Raharjo, R., Freeman, B.D. and Roman, I. (2006) Advanced Materials, 18, 39–44.

j83

5 Fundamentals in Electromembrane Separation Processes Heinrich Strathmann

5.1 Introduction

Electromembrane processes such as electrolysis and electrodialysis have experienced a steady growth since they made their ﬁrst appearance in industrial-scale applications about 50 years ago [1–3]. Currently desalination of brackish water and chlorine–alkaline electrolysis are still the dominant applications of these processes. But a number of new applications in the chemical and biochemical industry, in the production of high-quality industrial process water and in the treatment of industrial efﬂuents, have been identiﬁed more recently [4]. The development of processes such as continuous electrodeionization and the use of bipolar membranes have further extended the range of application of electromembrane processes far beyond their traditional use in water desalination and chlorine-alkaline production. The term electromembrane process is used to describe an entire family of processes that can be quite different in their basic concept and their application. However, they are all based on the same principle, which is the coupling of mass transport with an electrical current through an ion permselective membrane. Electromembrane processes can conveniently be divided into three types: (1) Electromembrane separation processes that are used to remove ionic components such as salts or acids and bases from electrolyte solutions due to an externally applied electrical potential gradient. (2) Electromembrane synthesis processes that are used to produce certain compounds such as NaOH, and Cl2 from NaCL due to an externally applied electrical potential and an electrochemical electrode reaction. (3) Eletectromembrane energy conversion processes that are to convert chemical into electrical energy, as in the H2/O2 fuel cell. In this chapter only electromenbrane separation processes such as electrodialysis, electrodialysis with bipolar membranes and continuous electrodeionization will be discussed.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 5 Fundamentals in Electromembrane Separation Processes

84

5.2 The Structures and Functions of Ion-Exchange Membranes

The key components in electrodialysis and related processes are the ion-exchange membranes. There are three different types of ion-exchange membranes: (1) cationexchange membranes that contain negatively charged groups ﬁxed to the polymer matrix, (2) anion-exchange membranes that contain positively charged groups ﬁxed to the polymer matrix, and bipolar membranes that are composed of an anion- and a cation-exchange layer laminated together. In a cation-exchange membrane, the ﬁxed negative charges are in electrical equilibrium with mobile cations in the interstices of the polymer as indicated in Figure 5.1, which shows schematically the structure of a cation-exchange membrane with negative charges ﬁxed to the polymer matrix, and mobile cations and anions. The mobile cations are referred to as counterions and the mobile anions that carry the same electrical charge as the polymer membrane that are more or less completely excluded from the membrane are referred to as co ions. Due to the exclusion of the co ions, a cation-exchange membrane is more or less impermeable to anions. Anionexchange membranes carry positive ﬁxed charges and exclude cations. Thus, they are more or less impermeable to cations. To what extent the co ions are excluded from an ion-exchange membrane depends on membranes as well as on solution properties. Bipolar membranes enhance the dissociation of water molecules into H þ and OH ions and are used in combination with monopolar membranes for the production of acids and bases from the corresponding salts [5]. The most desired properties of ion-exchange membranes are: high permselectivity, low electrical resistance, good mechanical and form stability, and high chemical and thermal stability. In addition to these properties bipolar membranes should have high catalytic water dissociation rates.

Figure 5.1 Schematic drawing illustrating the structure of a cation-exchange membrane.

5.2 The Structures and Functions of Ion-Exchange Membranes

5.2.1 Ion-Exchange Membrane Materials and Structures

Many of todays available membranes meet most of these requirements. In particular, the Naﬁon-type cation-exchange membrane has quite satisfactory properties for applications in the chlorine–alkaline electrolyses as well as in electrodialysis [6]. Anion-exchange membranes often show lower stability in strong alkaline solutions than cation-exchange membranes. The properties of ion-exchange membranes are determined by two parameters, that is, the basic material they are made from and the type and concentration of the ﬁxed ionic moiety. The basic material determines to a large extent the mechanical, chemical, and thermal stability of the membrane. Ion-exchange membranes are made today from hydrocarbon or partially halogenated hydrocarbon and perﬂuorocarbon polymers [7, 8]. The type and the concentration of the ﬁxed ionic charges determine the permselectivity and the electrical resistance of the membrane, but they also have a signiﬁcant effect on the mechanical properties of the membrane. The degree of swelling, especially, is effected by the type of the ﬁxed charges and their concentration. The following moieties are used as ﬁxed charges in cation-exchange membranes: SO 3

COO

PO2 3

PHO 2

AsO2 3

SeO 3:

In anion-exchange membranes ﬁxed charges may be: þ

þ

N H2 R N HR2

þ

N R3

þ

P R3

þ

S R2 :

ThesulfonicacidgroupiscompletelydissociatedovernearlytheentirepHrange,while the carboxylic acid group is virtually undissociated in the pH range <3. The quaternary ammonium group again is completely dissociated over the entire pH range, while the secondary ammonium group is only weakly dissociated. Accordingly, ion-exchange membranes are referred to as being weakly or strongly acidic or basic in character. Ion-exchange membranes can also be divided, according to their structure and preparation procedure, into homogeneous and heterogeneous membranes [4]. In homogeneous ion-exchange membranes the ﬁxed-charged groups are evenly distributed over the entire membrane polymer matrix. Homogeneous membranes can be produced, for example, by polymerization or polycondensation of functional monomers such as phenolsulfonic acid, or by functionalizing a polymer such as polysulfone dissolved in an appropriate solvent by sulfonation. Heterogeneous ion-exchange membranes have distinct macroscopic domains of ion-exchange resins in the matrix of an uncharged polymer. They can be produced by melting and pressing a dry ion-exchange resin with a polymer powder such as polyvinylchloride, or by dispersion of the ion-exchange resin in a polymer solution. 5.2.2 Preparation of Ion-Exchange Membranes

Ion-exchange membranes are ion-exchange resins in sheet form. There are, however, signiﬁcant differences between ion-exchange resins and membranes as far as the

j85

j 5 Fundamentals in Electromembrane Separation Processes

86

mechanical properties and especially the swelling behavior are concerned. Ionexchange resins are mechanically weak or tend to be brittle. Changes in the electrolyte concentration of an electrolyte in equilibrium with the ion-exchange resin may cause major changes in the water uptake and hence in swelling. These changes can not be tolerated in ion-exchange membranes that have to ﬁt an apparatus under very different electrolyte concentrations and temperatures. The most common solution to this problem is the preparation of a membrane with a backing of a stable reinforcing material that gives the necessary strength and dimensional stability. Preparation procedures for making ion-exchange membranes are described in great detail in the literature [8–10] and are quite different for heterogeneous and homogeneous membranes. 5.2.2.1 Preparation Procedure of Heterogeneous Ion-Exchange Membranes Ion-exchange membranes with a heterogeneous structure consist of ﬁne ion-exchange particles embedded in an inert binder polymer such as polyethylene, phenolic resins, or polyvinylchloride. Heterogeneous ion-exchange membranes are characterized by the discontinuous phase of the ion-exchange material. The efﬁcient transport of ions through a heterogeneous membrane requires either a contact between the ion-exchange particles or an ion-conducting solution between the particles. Heterogeneous ion-exchange membranes can easily be prepared by mixing an ion-exchange powder with a dry binder polymer and extrusion of sheets under the appropriate conditions of pressure and temperature or by dispersion of ion-exchange particles in a solution containing a dissolved ﬁlm-forming binder polymer, casting the mixture into a ﬁlm and then evaporating the solvent. Heterogeneous ion-exchange membranes with useful low electrical resistances contain more than 65% by weight of the ion-exchange particles. Membranes that contain signiﬁcantly less than 65 wt% ion-exchange particles have high electric resistance and membranes with signiﬁcantly more resin particles have poor mechanical strength. Furthermore, heterogeneous membranes develop water-ﬁlled interstices in the polymer matrix during the swelling process that affects both the mechanical properties as well as the permselectivity. The ion-exchange capacities of heterogeneous membranes are in the range of 1–2 equivalent per kilogram dry membrane and thus signiﬁcantly lower than that of homogeneous membranes, which is between 2 and 3 equivalent per kilogram dry resin. In general, heterogeneous ion-exchange membranes have higher electrical resistances and lower permselectivity than homogeneous membranes. 5.2.2.2 Preparation of Homogeneous Ion-Exchange Membranes Homogeneous ion-exchange membranes can be prepared by polymerization of monomers that contain a moiety that either is or can be made anionic or cationic, or by polymerization of a monomer that contains an anionic or a cationic moiety, or by introduction of anionic or cationic moieties into a polymer dissolved in a solvent by a chemical reaction, or grafting functional groups into a preformed polymer ﬁlm [10]. A method of preparing both cation- and anion-exchange membranes, which is used for the preparation of commercial cation-exchange membranes, is the poly-

5.2 The Structures and Functions of Ion-Exchange Membranes

merization of styrene and divinylbenzene and its subsequent sulfonation according to the following reaction scheme:

In a ﬁrst step styrene is partially polymerized and cross-linked with divinylbenzene and then in a second step sulfonated with concentrated sulfuric acid. The obtained membranes show high ion-exchange capacity and low electrical resistance. To increase the mechanical strength the membrane is cast on a support screen. A homogeneous anion-exchange membrane can be obtained by introducing a quaternary amine group into polystyrene by a chloromethylation procedure followed by an amination with a tertiary amine according to the following reaction scheme:

The membrane structures and their preparation described above are just two examples. There are many variations of the basic preparation procedure resulting in slightly different products. Instead of styrene, often substituted styrenes such as methylstyrene or phenyl-acetate are used instead of divinylbenzene monomers such as divinylacetylene or butadiene are used. One of the technically and commercially most important cation-exchange membranes developed in recent years is based on perﬂuorocarbon polymers. Membranes of this type have extreme chemical and thermal stability and they are the key component in the chlorine–alkaline electrolysis as well as in most of todays fuel cells. They are prepared by copolymerization of tetraﬂuoroethylene with perﬂuorovinylether having a carboxylic or sulfonic acid group at the end of a side chain. There are several variations of a general basic structure commercially available today [11]. The various preparation techniques are described in detail in the patent literature.

j87

j 5 Fundamentals in Electromembrane Separation Processes

88

Todays commercially available perﬂuorocarbon membranes have the following basic structure: CF2 CF2

k

CF CF2 OCF2

l

CF m O

(CF2 )nX

CF3 k = 5–8, l = 600–1200, m = 1–2, n = 1– 4, X = SO3 –, COO–

The synthesis of the perﬂuorocarbon membranes is rather complex and requires a multistep process. In addition to the various perﬂuorinated cation-exchange membranes an anionexchange membrane has also been developed. The anion-exchange membrane has similar chemical and thermal properties to the perﬂuorinated cation-exchange membrane. More recently cation-exchange membranes with good mechanical and chemical stability and well-controlled ion-exchange capacity are prepared by dissolving and casting a functionalized polymer such as sulfonated polysulfone, or sulfonated polyetheretherketone in an appropriate solvent, followed by casting the mixture into a ﬁlm and then evaporation of the solvent [12]. To obtain membranes with different ion-exchange capacity the sulfonated polyetheretherketone or polysulfone can be mixed with unsulfonated polymer in a solvent such as N-methylpyrrolidone. By changing the ratio of the sulfonated to unsulfonated polymer the ﬁxed-charge density can easily be adjusted to a desired value. The sulfonated polysulfone as well as polyetheretherketone can be cast as a ﬁlm on a screen. After the evaporation of the solvent a reinforced membrane with excellent chemical and mechanical stabilities and good electrochemical properties is obtained. The anion-exchange membrane based on polysulfone can be prepared by halomethylation of the backbone polymer and subsequent reaction with a tertiary amine 5.2.2.3 Special Property Membranes In addition to the monopolar membrane described above a large number of special property membranes are used in various applications such as low-fouling anionexchange membranes used in certain wastewater treatment applications or composite membranes with a thin layer of weakly dissociated carboxylic acid groups on the surface used in the chlorine–alkaline production, and bipolar membranes composed of a laminate of an anion- and a cation-exchange layer used in the production of protons and hydroxide ions to convert a salt in the corresponding acids and bases. The preparation techniques are described in detail in numerous publications [13–15].

5.3 Transport of Ions in Membranes and Solutions

The transport rate of a component in a membrane and a solution is determined by its concentration, its mobility in a given environment and by the driving force or forces

5.3 Transport of Ions in Membranes and Solutions

acting on the component. The concentration and mobility of a component are determined by its interaction with other components in its surrounding. The driving forces for the transport are gradients in the electrochemical potential. In electrolyte solutions the electrostatic forces must always be balanced, that is, the electroneutrality prevails on a macroscopic scale. For applying an electrical potential in an electrolyte solution two electron conductors must be in contact with an electrolyte. At the electrode/electrolyte interface the electron conductance is converted to an ionic conductance by an electrochemical reaction. 5.3.1 Electric Current and Ohms Law in Electrolyte Solutions

In electromembrane processes the anions move towards the anode where they are oxidized by releasing electrons to the electrode in an electrochemical reaction. Likewise, the positively charged cations move towards the cathode where they are reduced by receiving electrons from the electrode in an electrochemical reaction. Thus, the transport of ions in an electrolyte solution and ion-exchange membrane between electrodes results in a transport of electrical charges, that is, an electrical current which can be described by the same mathematical relation as the transport of electrons in a metallic conductor, that is, by Ohms law that is given by: U ¼ RI

ð5:1Þ

Here, U is the electrical potential between two electron sources, for example, between two electrodes, I is the electrical current between the electron sources, and R the electrical resistance. The resistance R is a function of the speciﬁc resistance of the material, the distance between the electron sources, and the cross-sectional area of the material through which the electric current passes. It is given by: R¼r

l q

ð5:2Þ

Here, R is the overall resistance, r is the speciﬁc resistance, l is the length, and q the cross-sectional area of the conducting material. The reversal of the resistance and of the speciﬁc resistance, respectively, is the conductivity and the speciﬁc conductivity, thus is: S¼

1 R

and k ¼

1 r

ð5:3Þ

Here, S is the conductivity and k the speciﬁc conductivity. The conductivity of electrons in metal conductors, however, is generally 3–5 orders of magnitude higher than that of ions in electrolyte solutions. Furthermore, the conductivity of metals is decreasing with increasing temperature, while the

j89

j 5 Fundamentals in Electromembrane Separation Processes

90

conductivity of electrolyte solutions is increasing with temperature. The most important difference between electron and ion conductivity, however, is the fact that ion conductivity is always coupled with a transport of mass while, due to the very small mass of an electron, virtually no mass is transported in an electron conductor. The conductivity of electrolyte solutions depends on the concentration and the charge number of the ions in the solution. It is expressed as the molar or equivalent conductivity or molar conductivity, which is given by: Lmol ¼

k Cs

and

Leq ¼

k ðza na þ zc nc Þ Cs 2

ð5:4Þ

Here, Lmol and Leq are the molar and the equivalent conductivity, C is the molar concentration of the electrolyte in the solution, za and zc are the charge numbers of the anion and cation, respectively, and na and nc are the stoichoimetric coefﬁcients of the anion and cation, respectively. The stoichoimetric coefﬁcient gives the number of anions and cations in a mole electrolyte and the charge number gives the number of charges related to an ion. For example, for NaCl nc and na are identical and 1 and also za and zc are 1. However, for MgCl2 nc is 1 and na is 2, and zc is 2 and za is 1. The number of electrical charges carried by all the ions of an electrolyte under the driving force of an electrical potential gradient through a certain area A in the direction of transport is given by: Je ¼

X i

zi ui ni C F

X Df X Df ¼ zi FJi ¼ zi ni Cleq Dz Dz i i

ð5:5Þ

Here, Je is the ﬂux of electrical charges and Ji that of the individual ions, z, u, and n are the charge number, the ion mobility, and the stoichiometric coefﬁcient, respectively, C is the concentration of the electrolyte, Dj and Dz are the potential difference and the distance between two points in the z direction, F is the Faraday constant, which is F ¼ 96 485 [C eq1], and leq is the equivalent conductivity. Thus, the ﬂux of electrical charges represents an electrical current, which is according to Ohms law given by: I¼

X U X Dj Dj ¼ A¼k ¼ iA zi FJi A ¼ zi ni Cleq R l l i i

ð5:6Þ

Here, I is the current, U is the applied voltage, R is the resistance, A is the area through which the current passes, Dj is the voltage difference between two points, and l is the distance between the two points, k is the conductivity, A is the crosssectional area of the conducting media, and i is the current density

5.3 Transport of Ions in Membranes and Solutions

5.3.2 Mass Transport in Membranes and Solutions

To describe the mass transport in an electrolyte solution or in an ion-exchange membrane, three independent ﬂuxes must be considered, that is, the ﬂuxes of the cations the ﬂux of anions, and the ﬂux of the solvent [16]. The transport of ions is the result of an electrochemical potential gradient and the transport of the solvent through the membrane is a result of osmotic and electro-osmotic effects. 5.3.2.1 The Driving Force and Fluxes in Electromembrane Processes The driving force for the ﬂux of a component in electromembrane processes is a gradient in their electrochemical potential which is given at constant temperature by:

~ i dmi df dm i dp þ RT d ln ai þ zi F df ¼ þ ¼V dz dz dz dz dz dz

ð5:7Þ

~ i , dhi, dmi, d ln ai, dj and dp are the gradients of the electrochemical potential, Here dm the chemical potential, the activity, the electrical potential and of the hydrostatic pressure, F is the Faraday constant and R the gas constant, and T the temperature. The mass transport in electromembrane processes at constant pressure and temperature can be described as a function of the driving force by a phenomenological equation [17], that is,: X d ln ai X dm ~ df ð5:8Þ þ zi F Lik k ¼ Lik RT Ji ¼ dz dz dz i i Here, Lik is a phenomenological coefﬁcient relating the driving force to the corresponding ﬂux, the subscripts i and k refer to various components in the system. Assuming an ideal solution in which the activity of a component is identical to its concentration and no kinetic coupling occurs between individual ﬂuxes, Equation 5.8 becomes identical with the Nernst–Planck ﬂux equation [18], which is given by: dCi zi FCi dj þ ð5:9Þ Ji ¼ Di dz RT dz Here, Di is the diffusion coefﬁcient of the component i which is related to the phenomenological coefﬁcient by: Di ¼ (Lii/RT). The ﬁrst term Di(dCi/dz) represents the diffusion, the second term Di(ziCiF/RT) (dj/dz) the migration of a component. Thus, the Nernst–Planck equation is an approximation of the more general phenomenological equation. 5.3.2.2 Electrical Current and Fluxes of Ions The electric current in an electrolyte solution is transported by ions only. as described in Equation 5.6:

i¼

X I ¼F z i Ji A i

ð5:10Þ

j91

j 5 Fundamentals in Electromembrane Separation Processes

92

Here, i is the current density, I the current, A the membrane surface, F the Faraday constant, J the ﬂux, and z the valence, the subscript i refers to cations and anions. Introducing Equation 5.8 and 5.9 and rearranging leads to: X Ci Di RT dCi dj X þ ð5:11Þ i¼F z i Ji ¼ F 2 z2i dz RT zi Ci F dz i i Here, i is the current density, C is the concentration, F is the Faraday constant, j is the electrical potential, z is the valence, D is the ion diffusivity, R is the gas constant, T is the absolute temperature, and the subscript i refers to anions and cations. The term (RT/ziCiF)(dCi/dz) has the dimensions of an electrical potential gradient and represents the concentration potential that is established between two electrolyte solutions of different concentrations. 5.3.2.3 The Transport Number and the Membrane Permselectivity In an electrolyte solution the current is carried by both ions. However, cations and anions usually carry different portions of the overall current. In ion-exchange membranes the current is carried preferentially by the counterions. The fraction of the current that is carried by a certain ion is expressed by the ion transport number, which is given by:

z i Ji Ti ¼ P z i Ji

ð5:12Þ

i

Here, Ti is the transport number of the component i, Ji is its ﬂux, and zi its valence. The transport number Ti indicates the fraction of the total current that is carried by the ion i, the sum of the transport number of all ions in a solution is 1. The membrane permselectivity is an important parameter for determining the performance of a membrane in a certain ion-exchange membrane process. It describes the degree to which a membrane passes an ion of one charge and retains an ion of the opposite charge. The permselectivity of cation- and anion-exchange membranes can be deﬁned by the following relations [4]: Ycm ¼

Tccm Tc Ta

and

Yam ¼

Taam Ta Tc

ð5:13Þ

Here, Y is the permselectivity of a membrane, T is the transport number, the superscripts cm and am refer to cation- and anion-exchange membranes, and the subscripts c and a refer to cation and anion, respectively. An ideal permselective cation-exchange membrane would transmit positively charged ions only, that is, for a transport number of a counterion in a cation-exchange membrane is Tccm ¼ 1 and the permselectivity of the membrane is Ycm ¼ 1. The permselectivity approaches zero when the transport number within the membrane is identical to that in the electrolyte solution, that is, for Tccm ¼ Tc is Ycm ¼ 0. For the anion-exchange membrane the corresponding relation holds. The transport number of a certain ion in the membrane is proportional to its concentration in the membrane that again is a function of its concentration in the

5.3 Transport of Ions in Membranes and Solutions

solutions in equilibrium with the membrane phase, due to the Donnan exclusion. Thus, the selectivity of ion-exchange membranes results from the exclusion of co ions from the membrane phase. The concentration of a co ion in an ion-exchange membrane can be calculated from the Donnan equilibrium. For a monovalent salt and a dilute salt solution and assuming the activity coefﬁcients of the salt in the membrane and the solution to be 1, the co ion concentration in the membrane is given to a ﬁrst approximation by: m

Cco ¼

s

C2s Cfix

ð5:14Þ

Here C is the concentration, the subscripts co, s and ﬁx refer to co ion, salt and ﬁxed ion of the membrane, the superscripts s and m refer to membrane and solution. Equation 5.14 indicates that the co ion concentration in the membrane and with that the permselectivity of the membrane is decreasing with salt concentration in the solution and will vanish when the salt concentration in the solution is identical to the ﬁxed ion concentration of the membrane. 5.3.2.4 Membrane Counterion Permselectivity The transport number of counterions in an ion-exchange membrane is always quite high compared to that of the co-ionco-ions. But the transport number of different counterions can be quite different, too. The transport rates of ions in a solution or through a membrane are determined by their concentration and mobility in the membrane. The concentration of the counterions is always close to the concentration of the ﬁxed charges of the membrane. The mobility of the ions in the membrane depends mainly on the radius of the hydrated ions and the membrane structure. The mobility of different ions in an aqueous solution does not differ very much from each other. An exception is the H þ and OH ions. Their mobility is about a factor 5 to 8 higher than that of other ions. This exceptionally high mobility of the H þ ion can be explained by the transport mechanism of protons and hydroxide ions. Because of the molecular interaction of water dipoles with electrical charges, protons form hydronium ions. Common salt ions move with their hydrate shell through the solution. The proton, however, is transported mostly via a so-called tunnel mechanism from one hydronium ion to the next water molecule. This explains not only the extraordinarily high mobility of protons but it is also one of the reasons for the high permeability of anion-exchange membranes for protons, while these membranes generally have a very low permeability for salt cations. The same mechanism also holds true for the transport of hydroxide ions and thus the permeability of hydroxide ions in an aqueous solution and also in a cation-exchange membrane is much higher than that of other salt anions. Because protons and hydroxide ions are transported only to a small extent as individual ions surrounded by a hydration shell they contribute very little to the electro-osmotic transport of water, and their water-transport number is always quite low. The permselectivity of an ion-exchange membrane for different counterions is determined by the concentration and the mobility of the different ions in the membrane as indicated earlier. The concentration of the different counterions in

j93

j 5 Fundamentals in Electromembrane Separation Processes

94

the membrane is determined mainly by electrostatic effects referred to as electroselectivity [9]. The mobility depends on the size of the hydrated ion. A typical counterion-exchange sequence of a cation-exchange membrane containing SO3 group as ﬁxed charge is: þ þ Ba2þ > Pb2þ > Sr2þ > Ca2þ > Mg2þ > Agþ > Kþ > NHþ 4 > Na > Li

A similar counterion-exchange sequence is obtained for anions in an anionexchange membrane containing quaternary ammonium groups as ﬁxed charges: 2 I > NO 3 > Br > Cl > SO4 > F

The permselectivity is the product of ion-exchange selectivity and mobility selectivity. The mobility of different ions is determined mainly by steric effects, that is, the size of the ions and the cross-linking density of the membrane [4]. 5.3.2.5 Water Transport in Electrodialysis Water transport in electrodialysis from the diluate to the concentrate process stream can affect the process efﬁciency signiﬁcantly. If a convective ﬂux as a result of pressure differences between ﬂow streams can be excluded there are still two sources for the transport of water from the diluate to the concentrate solution. The ﬁrst one is the result of osmotic-pressure differences between the two solutions, and the second is due to electro-osmosis that results from the coupling of water to the ions being transported through the membrane due to the driving force of an electrical potential. Each of the two ﬂuxes may be dominant depending on the permselectivity of the ion-exchange membrane, the concentration gradient, and the current density. In a highly permselective membrane and with moderate differences in the salt concentration in the two solutions separated by the membrane the electro-osmotic ﬂux is dominating and generally much higher than the osmotic solvent ﬂux. In electrodialysis the water ﬂux due to electro-osmosis can be expressed by a solvent transport number which gives the number of water molecules transported by one ion: X Ji ð5:15Þ Jw ¼ m T w i

Here, T w is the water transport number, Jw is the water ﬂux, and Ji is the ﬂux of ions through a given membrane. The water transport number thus is: m

m

Jw Tw ¼ X

Ji

ð5:16Þ

i

The water-transport number refers to the number of water molecules transferred by one ion through a given membrane. It depends on the membrane and on the electrolyte, that is, on the size of the ions, their valence, and their concentration in the solution. In aqueous salt solutions and commercial ion-exchange membranes the water transport number is of the order of 4–8, that is, one mole of ions transports about 4–8 moles of water through a typical commercial ion-exchange membrane.

5.4 The Principle of Electromembrane Processes

5.4 The Principle of Electromembrane Processes

In this chapter only electromembrane separation processes such as electrodialysis, electrodialysis with bipolar membranes, and continuous electrodeionization will be discussed. 5.4.1 Electrodialysis

Electrodialysis is the most important electromembrane process and one of the ﬁrst membrane processes used for desalination of brackish water to produce high-quality potable water at acceptable costs on a large commercial scale. Today, the process has found a multitude of applications in preconcentration of seawater for the production of table salt or in recovering valuable constituents from industrial efﬂuents [19]. The principle of electrodialysis is illustrated in Figure 5.2 which shows a schematic diagram of an electrodialysis stack consisting of a series of anion- and cationexchange membranes arranged in an alternating pattern to form individual cells between an anode and a cathode. If an ionic solution such as an aqueous salt solution is pumped through these cells and an electrical potential is established between the anode and cathode, the positively charged cations migrate towards the cathode and the negatively charged anions towards the anode. The cations permeate the cationexchange membrane but are retained by the anion-exchange membrane. Likewise, the negatively charged anions permeate the anion-exchange membrane and are retained by the cation-exchange membrane. The overall result is an increase in the

Figure 5.2 Schematic diagram illustrating the principle of desalination by electrodialysis in a stack with cation- and anionexchange membranes in alternating series between two electrodes.

j95

j 5 Fundamentals in Electromembrane Separation Processes

96

ion concentration in alternate compartments, while the other compartments simultaneously become depleted. The depleted solution is generally referred to as the diluate and the concentrated solution as the brine or the concentrate. The driving force for the ion transport in the electrodialysis process is the applied electrical potential between the anode and cathode. The total space occupied by the diluate and the concentrated solution and the anion- and cation-exchange membranes separating the solutions make up a cell pair that represents a repeating unit between the electrodes. 5.4.1.1 Electrodialysis System and Process Design The efﬁciency of electrodialysis is determined to a large extent by the properties of the membranes. But it is also affected by the process and system design that determine the limiting current density, the current utilization, the concentration polarization and the overall efﬁciency and costs [20, 21]. The electrodialysis stack A key element in electrodialysis is the so-called stack, which is a device to hold an array of membranes between the electrodes that the streams being processed are kept separated. A typical electrodialysis stack used in water desalination contains 100–300 cell pairs stacked between the electrodes. The electrode containing cells at both ends of a stack are often rinsed with a separate solution which does not contain Cl ions to avoid chlorine formation. The membranes in an electrodialysis cell are separated by spacer gaskets as indicated in Figure 5.3, which shows schematically the design of a so-called sheet ﬂow electrodialysis stack. The spacer gasket consists of a screen that supports the membranes and controls the ﬂow distribution in the cell and a gasket that seals the cell to the outside and also contains the manifolds to distribute the process ﬂuids in

Figure 5.3 Exploded view of a sheet-flow-type electrodialysis stack arrangement indicating the individual cells and the spacer gaskets containing the manifold for the distribution of the different flow streams.

5.4 The Principle of Electromembrane Processes

the different compartments. To minimize the resistance of the solution in the cell the distance between two membranes is kept as small as possible and is in the range of 0.5–2 mm in industrial electrodialysis stacks. A proper electrodialysis stack design provides a maximum effective membrane area per unit stack volume and ensures uniform ﬂow distribution and mixing of the solutions to minimize concentration polarization at the membrane surfaces, but also minimizes the pressure loss of the solution ﬂow in the stack. concentration polarization and limiting current density The limiting current density is the maximum current that may pass through a given cell pair area without detrimental effects. If the limiting current density is exceeded, the electric resistance in the diluate will increase and water dissociation may occur at the membrane surface that can lead to pH changes in the solutions and effect the current utilization. The limiting current density is determined by concentration-polarization effects at the membrane surface in the diluate containing compartment that in turn is determined by the diluate concentration, the compartment design, and the feedﬂow velocity. Concentration polarization in electrodialysis is also the result of differences in the transport number of ions in the solution and in the membrane. The transport number of a counterion in an ion-exchange membrane is generally close to 1 and that of the co ion close to 0, while in the solution the transport numbers of anion and cations are not very different. At the surface of a cation-exchange membrane facing the diluate solution the concentration of ions in the solution is reduced because of the lower transport number of the cations in the solution than in the membrane. Because of the electroneutrality requirements the number of anions is reduced in the boundary layer by migration in the opposite direction. The net result is a reduction of the electrolyte concentration in the solution at the surface of the membrane and a concentration gradient is established in the solution between the membrane surface and the well-mixed bulk. This concentration gradient results in a diffusive electrolyte transport. A steady-state situation is obtained when the additional ions that are needed to balance those removed from the interface due to the faster transport rate in the membrane are supplied by the diffusive transport. The other side of the cation-exchange membrane is facing the concentrate solution and here the opposite effect occurs and the electrolyte concentration at the membrane surface is increased accordingly. The concentration polarization is limited to the laminar boundary layer at the membrane surface, which is very thin due to turbulent mixing of the bulk solution. The effect of concentration polarization is illustrated in Figure 5.4 which shows the salt concentration proﬁles and the ﬂuxes of cations and anions in the concentrate and diluate solution at the surface of a cation-exchange membrane. The symbols J and C in Figure 5.5 denote the ﬂuxes and the concentration of ions, the superscripts mig and diff refer to migration and diffusion, the superscripts d and c refer to diluate and concentrate solution, and the superscripts b and m refer to bulk phase and membrane surface, respectively, the subscripts a and c refer to anion and cation.

j97

j 5 Fundamentals in Electromembrane Separation Processes

98

Figure 5.4 Schematic drawing illustrating the concentration profiles of a salt in the laminar boundary layer on both sides of a cation-exchange membrane and the flux of ions in the solutions and the membrane.

The concentration polarization occurring in electrodialysis, that is, the concentration proﬁles at the membrane surface can be calculated by a mass balance taking into account all ﬂuxes in the boundary layer and the hydrodynamic conditions in the ﬂow channel between the membranes. To a ﬁrst approximation the salt concentration at the membrane surface can be calculated and related to the current density by applying the so-called Nernst ﬁlm model, which assumes that the bulk solution between the laminar boundary layers has a uniform concentration, whereas the concentration in the boundary layers changes over the thickness of the boundary layer. However, the concentration at the membrane surface and the boundary layer thickness are constant along the ﬂow channel from the cell entrance to the exit. In a practical electrodialysis stack there will be entrance and exit effects and concentration

Figure 5.5 Schematic drawing illustrating the sheet-flow and a tortuous-path spacer concept.

5.4 The Principle of Electromembrane Processes

differences between the solutions in the entrance and exit region of the cell, and the idealized model hardly exists. Nevertheless, the Nernst model provides a very simple approach to the mathematical treatment of the concentration polarization, which results in an expression for the current density as a function of the bulk solution concentration, the transport number of the ions, the diffusion coefﬁcient of the electrolyte and the thickness of the laminar boundary layer [20]. i¼

zi FDi DCid ðTim Tisi Þ Dz

ð5:17Þ

Here, T is the transport number of the counterion, DC is the concentration difference between the solution in the diluate at the membrane surface and in the bulk, D is the diffusion coefﬁcient, T is the transport number, F is the Faraday constant, z is the charge number, and Dz is the boundary layer thickness, the subscript i refers to cations or anions; the superscripts d, m and s refer to diluate, membrane and solution, respectively. When the ﬂow conditions are kept constant the boundary layer will be constant and the current density will reach a maximum value independent of the applied electrical potential gradient if the counterion concentration and thus the salt concentration at the membrane surface become 0. The maximum current density is referred to as the limiting current density. Thus is i ¼ ilim for m Cds ! 0 and b d zi FDs Cs ð5:18Þ ilim ¼ m ðTi Ti Þ Dz Here, ilim is the limiting current density, b Cds is the salt concentration of the diluate in the bulk solution, Dz is the thickness of the laminar boundary layer, T m and T s are the transport numbers in the membrane and the solution, Ds is the salt diffusion coefﬁcient in the solution, F is the Faraday constant, z is the charge number, and the subscript i refers to cation and anion. Exceeding the limiting current density in practical applications of electrodialysis can affect the efﬁciency of the process severely by increasing the electrical resistance of the solution and causing water dissociation, which leads to changes of the pH values of the solution causing precipitation of metal hydroxide on the membrane surface. Since the thickness of the laminar boundary in an electrodialysis stack is difﬁcult to determine in an independent measurement, the limiting current density in practical application is generally not calculated by Equation 5.18 but by an experimentally determined relation which describes the limiting current density as a function to the feed-ﬂow velocity in the electrodialysis stack [4]. The limiting current density is expressed by: ilim ¼ a ub FCsd

ð5:19Þ

Here, Csd is the concentration of the solution in the diluate cell, u is the linear ﬂow velocity of the solution through the cells parallel to the membrane surface, F is the Faraday constant, and a and b are characteristic constants for a given stack design and must be determined experimentally. This is done in practice by measuring the

j99

j 5 Fundamentals in Electromembrane Separation Processes

100

limiting current density in a given stack conﬁguration and constant feed solution salt concentrations as a function of the feed-ﬂow velocity. Current utilization In practical application electrodialysis is affected by incomplete current utilization. The reasons for the incomplete current utilization are poor membrane permselectivity, parallel current through the stack manifold, and water transport by convection and due to osmosis and electro-osmosis. In a well-designed stack with no pressure difference between diluate and the concentrate convective water transport is negligibly low and also the current through the manifold can be neglected. Under these conditions the overall current utilization is given by: c w Cs Csd ð5:20Þ x ¼ n ycm Tas þ yam Tcs 1 Twcm þ Twam V

Here, x is the current utilization, y is the membrane permselectivity, T is the w is the partial molar transport number, n is the number of cell pairs in the stack, V volume of water, and C is the concentration, a, c, s and w refer to anion, cation, solution and water, respectively, and the superscripts cm, am, c, and d refer to cationexchange membrane, anion-exchange membrane, concentrate and diluate. Electrodialysis equipment and process design The performance of electrodialysis in practical applications is not only a function of membrane properties but is also determined by the equipment and overall process design. As far as the stack design is concerned there are two major concepts used on a large scale. One is the sheet-ﬂow concept, which is illustrated in Figure 5.3 and the other is the so-called tortuous path concept, which is illustrated in Figure 5.5. The main difference between the sheet-ﬂow and the tortuous-path ﬂow spacer is that in the sheet-ﬂow spacer the compartments are vertically arranged and the process path is relative short. The ﬂow velocity of the feed is between 2 and 4 cm/s and the pressure loss correspondingly low, that is, between 0.2 and 0.4 bars. In the tortuous-path ﬂow stack, the membrane spacers are horizontally arranged and have a long serpentine cut-out that deﬁnes a long narrow channel for the ﬂuid path. The feed-ﬂow velocity in the stack is relatively high, that is, between 6 and 12 cm/s, which provides a better control of concentration polarization and higher limiting current densities, but the pressure loss in the feed-ﬂow channels is quite high, that is, between1 and 2 bars. However, higher velocities help to reduce the deposition of suspended solids such as polyelectrolytes, humic acids, surfactants, and biological materials on the membrane surface. In the practical application of electrodialysis there are two main process operation modes. The ﬁrst one is referred to as the unidirectional electrodialysis and the second as electrodialysis reversal [22]. In a unidirectional operated electrodialysis system the electric ﬁeld is permanently applied in one direction and the diluate and concentrate cells are also permanently ﬁxed over the period of operation. Unidirectional operated electrodialysis plants are rather sensitive to membrane fouling and scaling and often require a substantial feed-solution pretreatment and stack-cleaning procedures in the form of periodical rinsing of the stack with acid or detergent solutions. The unidirectional operating concept is mainly used today for applications in the

5.4 The Principle of Electromembrane Processes

Figure 5.6 Schematic drawing illustrating the removal of deposited negatively charged colloidal components from the surface of an anion-exchange membrane by reversing the electric field.

food and drug industry where often solutions contain valuable components that must be recovered in the concentrate or in the diluate. In desalination of brackish or surface waters generally electrodialysis reversal is applied, which always results in some loss of the product water. In the electrodialysis reversal operating mode the polarity of the electric ﬁeld applied to the electrodialysis stack is reversed in certain time intervals. Simultaneously the ﬂow streams are reversed, that is, the diluate cell becomes the concentrate cell and vice versa with the result that matter being precipitated at the membrane surface will be redisolved and removed with the ﬂow stream passing through the cell [22]. The principle of the electrodialysis reversal operating mode is illustrated in Figure 5.6 that shows an electrodialysis cell formed by a cation- and anion-exchange membrane between two electrodes. If an electric ﬁeld is applied to a feed solution containing negatively charged particles or large organic anions these components will migrate to the anion-exchange membrane and be deposited on its surface to form a so-called fouling layer that can increase the resistance of the membrane dramatically. If the polarity is reversed the negatively charged components will now migrate away from the anion-exchange membrane back into the feed stream and the membrane properties are restored. This procedure has been very effective not only for the removal of precipitated colloidal materials but also for removing precipitated salts and is used today in almost all electrodialysis water-desalination systems. However, reversing the polarity of a stack has to be accompanied with a reversal of the ﬂow streams. This always leads to some loss of product and requires a more sophisticated ﬂow control. The ﬂow scheme of an electrodialysis plant operated with reversed polarity is shown in Figure 5.7. In the reverse-polarity operating mode, the hydraulic ﬂow streams are reversed simultaneously, that is, the diluate cell will become the brine cell and vice versa. In this operating mode, the polarity of the current is changed at speciﬁc time intervals ranging from a few minutes to several hours.

j101

j 5 Fundamentals in Electromembrane Separation Processes

102

Figure 5.7 Flow scheme of electrodialysis reversal in a continuous operating mode with the feed solution also used as electrode rinse.

The advantage of the reverse-polarity operating mode is that precipitation in the brine cells will be re-dissolved when the brine cell becomes the diluate cell in the reverse operating mode. During the reversal of the polarity and the ﬂow streams, there is a brief period when the concentration of the desalted product exceeds the product quality speciﬁcation. The product water outlet has a concentration sensor that controls an additional three-way valve. This valve diverts highly concentrated product to waste and then, when the concentration returns to the speciﬁed quality, directs the ﬂow to the product outlet. Thus, in electrodialysis reversal there is always a certain amount of the product lost to the waste stream. This is generally no problem in desalination of brackish water. It might, however, be not acceptable in certain applications in the food and drug industry when feed solutions with high value products are processed. The degree of desalination that can be achieved in passing the feed solution through a stack is a function of the solution concentration, the applied current density, and the residence time of the solution in the stack. If the ﬂow rates of diluate and concentrate through the stack are relatively high the degree of desalination or concentration that can be achieved in a single path is quite low and often not sufﬁcient to meet the required product qualities. If this is the case the electrodialysis can be operated as a process with feed and bleed in which the diluate or the concentrate or both are partially recycled as shown in Figure 5.8. In the feed and bleed mode both the brine and the product concentration can be determined independently and very high recovery rates can be obtained. 5.4.1.2 Electrodialysis Process Costs The total costs in electrodialysis are the sum of ﬁxed charges associated with the amortization of the plant capital costs and the plant operating costs. Both the capital costs as well as the plant operating costs per unit product are proportional to the number of ions removed from a feed solution, that is, the concentration difference

5.4 The Principle of Electromembrane Processes

Figure 5.8 Flow scheme of an electrodialysis stack operated in a feed and bleed mode, that is, with partial recycling of the diluate and concentrate solutions.

between the feed and the product solution. But they are also strongly affected by the plant capacity and location and the composition of the feed water and the overall process design [23]. Capital-related costs The capital costs are determined mainly by the required membrane area for a certain plant capacity and feed and required product concentration. Other items such as pumps and process control equipment are considered as a fraction of the required membrane area. This fraction depends on the plant capacity. The same is true for the required land that also depends on the location of the plant. The required membrane area for a given capacity plant can be calculated from the current density in a stack that again depends on feed and product solution concentration. It can be calculated for a solution containing a single monovalent salt such as NaCl from the total current passing through the stack which is given by:

I ¼

d Qcell FðC f Cd Þ ¼ Ai xcell

ð5:21Þ

d Qcell FðCf Cd Þ ixcell

ð5:22Þ

Thus: A¼

Here, I and i are the electric current and the current density passing through a cell pair, A is the cell area, Q is the volume ﬂow, C is the concentration expressed in equivalent per volume, F is the Faraday constant, and x the current utilization. The subscripts cell refers to the diluate cell, and the superscripts d and f refer to diluate and feed solution, respectively. The voltage drop across a cell pair is constant over the entire length of a cell pair from the feed entrance to the product exit while the resistance of the cell pair is changing from the feed inlet to the product exit due to a decrease of the resistance of the diluate concentration. Therefore, the current density is also decreasing along the length of a cell pair. The current density is related to the resistance and the voltage by: U ð5:23Þ i¼ RA is the average resistance in a cell Here, U is the voltage drop across a cell pair A and R pair.

j103

j 5 Fundamentals in Electromembrane Separation Processes

104

The voltage drop in an electrodialysis cell pair is the result of the resistances of the membranes and the solutions and the concentration potential between the concentrate and diluate, which generally can be neglected. Most electrodialysis stacks used in practical applications consist of geometrically identical cells that are operated in cocurrent ﬂow [23]. If, furthermore, it is assumed that to a ﬁrst approximation the conductivity is independent of the concentration in the range of interest and the salt activity can be replaced by the concentration the voltage drop at any point for a single mono-valent salt solution across a cell pair length is given by. D 1 1 am cm þr þr ð5:24Þ þ U¼i L Cd Cc Here, D is the cell thickness, L is the equivalent conductivity of the salt solution, r is the area resistance, the superscripts d, c, am and cm refer to the diluate, the concentrate, and the anion- and the cation-exchange membranes, respectively. The electrical resistance at any point along the cell length is given by: U 1 D 1 1 am cm R¼ ¼ þ r ð5:25Þ þ þ r I A L Cd Cc The average resistance over the entire length of the ﬂow channel is determined by the integral average of the solution concentrations. Thus is: 2 3 Cfd C c 6 Dln Cfc C d 7 ¼ 16 R ð5:26Þ þ r am þ r cm 7 4 5 fd d A LðC C Þ is the average resistance and A the area of a cell pair, Cfd and Cd are the salt Here, R concentrations of the diluate at the inlet and outlet of the cell, Cfc and Cc are the salt concentrations of the concentrate cell at the inlet and outlet, ram and rcm are the area resistances of the anion- and cation-exchange membranes. The membrane area required for a certain plant capacity as a function of the feed and product concentration of a single mono-valent salt is obtained by combination of Equations 5.22–5.26 and rearranging: " # Cfd Cc Lðr am þ r cm Þ Cfd C d ln fc d þ d D C C Qcell F Cd ð5:27Þ Atot ¼ Ncell d ilim xcell C LC d am cm r þ 1þ ð þ r Þ Cc D Here Atot is the total membrane area in a stack and Ncell is the number of cell pairs in a stack and ilim is the limiting current density that determines the maximum voltage that can be applied. All other symbols are identical as the ones in the Equations 5.21–5.27. The total investment-related costs depend on the price of the membranes and their useful life under operating conditions, which is in practical application 5–8 years, and on the price of the additional plant components and their life.

5.4 The Principle of Electromembrane Processes

Operating costs The operating costs are composed of labor cost, the maintenance of the plant and energy costs. The labor costs are general directly proportional to the size of the plant and usually calculated as a certain percentage of the investment related costs. The energy required in an electrodialysis process is an additive of two terms: (1) the electrical energy to transfer the ionic components from one solution through membranes into another solution and (2) the energy required to pump the solutions through the electrodialysis unit. Depending on various process parameters, particularly the feed-solution concentration, either one of the two terms may be dominating, thus determining the overall energy costs. The energy consumption due to electrode reactions can generally be neglected since more than 200 cell pairs are placed between the two electrodes in a modern electrodialysis stack. The energy required for operating the process control devices can be neglected. The total energy required in electrodialysis for the actual desalination process is given by the current passing through the electrodialysis stack multiplied with the total voltage drop encountered between the electrodes:

Edes ¼ Ist Ust t ¼ Ist Ncell Ucell t ¼ I 2 Ncell Rt

ð5:28Þ

Here, Edes is the energy consumed in a stack for the transfer of ions from a feed to a concentrate solution, Ist is the current passing through the stack, Ust and Ucell are the voltage applied across the stack, that is, between the electrodes, and across a cell pair; t is the time of operation. The total current through the stack is given by Equation 5.21 and the average resistance is given by Equation 5.26. Combination of the two Equations and multiplication by the number of cell pairs in the stack gives the desalination energy: 2 3 Cfd Cc d Dln fd d 2 6 Cfc Cd þ r am þ r cm 7 7 Qcell FðCs Cs Þ cell I 2 t ¼ Ncell t 6 Edes ¼ Ncell R 5 A 4LðCfd Cd Þ x ð5:29Þ The speciﬁc desalination energy, that is, the energy used per unit product volume is given by: 2 3 Cfd Cc d Dln fd d 2 2 cell I t Ncell t 6 Ncell R Cfc Cd þ r am þ r cm 7 6 7 Qcell FðCs Cs Þ ¼ Ede;spc ¼ 5 x Vpro AVpro 4LðCfd Cd Þ ð5:30Þ Here, Edes and Edes,spc are the desalination energy and the speciﬁc desalination, I is the total current, t is the time of operation; Cfd and Cfc are the equivalent concentrations of the diluate and the concentrate at the cell inlet, Cd and Cc are the concentrations of the diluate and the concentrate at the cell outlet, L is the equivalent conductivity of the salt solution, ram and rcm are the area resistances of the anion- and cation-exchange d is the diluate membrane, D is the cell thickness, x is the current utilization, and Qcell

j105

j 5 Fundamentals in Electromembrane Separation Processes

106

Figure 5.9 Schematic diagram illustrating the various cost items in electrodialysis as a function of the applied current density.

is the average resistance of a cell pair, A is the cell pair area, Ncell is ﬂow rate in a cell, R the number of cell pairs in a stack, and Vpro is a volume product water. Equation 5.30 shows that the energy dissipation due to the resistance of the solutions and membranes is increasing with the current density, since the electrical energy for a given resistance is proportional to the square of the current, whereas the salt transfer is directly proportional to the current. Hence, the power necessary for the production of a given amount of product increases with the current density. The higher the current density the more power is needed to maintain a given production rate. However, the higher the current density the lower is the required membrane area for a given capacity installation, as illustrated in Figure 5.9, which shows the total costs of desalination and the membrane area and current density related costs as a function of the current density. Figure 5.9 shows that at a certain current density and in the installed membrane area the total desalination costs reach a minimum that must be experimentally determined for a given stack design and feed, diluate, and concentrate. However, the upper limit for the current density of any given installation is determined by the limiting current density that should not be exceeded. The operation of an electrodialysis unit requires one or more pumps to circulate the diluate, the concentrate, and the electrode rinse solution through the stack. The energy required for pumping these solutions is determined by the volumes of the solutions to be pumped and the pressure drop. It can be expressed by: Ep;spe ¼

Ep ðQ d Dpd þ Q c Dpc þ Q e Dpe Þ ¼ k eff Q dt Qd

ð5:31Þ

Here, Ep,spec is the total energy for pumping the diluate, the concentrate, and the electrode rinse solution through the stack per unit diluate water, keff is an efﬁciency term for the pumps, Qd, Qc, and Qe are the volume ﬂow rates of the diluate, the concentrate, and the electrode rinse solution through the stack. The energy consumption due to the pressure loss in the electrode rinse solution can be neglected in most practical applications because the volume of the electrode rinse solution is very small compared to the volumes of the diluate and concentrate. The pressure losses in the various cells are determined by the solution ﬂow velocities and the cell design. The energy requirements for circulating the solution

5.4 The Principle of Electromembrane Processes

Figure 5.10 Schematic diagram illustrating the acid and base production from the corresponding salt by electrodialysis with bipolar membranes.

through the system may become a signiﬁcant or even dominant portion of the total energy consumption for solutions with rather low salt concentration. 5.4.2 Electrodialysis with Bipolar Membranes

Electrodialysis with bipolar membranes has gained increasing attention as an efﬁcient tool for the production of acids and bases from the corresponding salts. This process is economically attractive and has many potential applications [24, 25]. A typical arrangement of an electrodialysis stack with bipolar membranes is illustrated in Figure 5.10, which shows the production of an acid and a base from the corresponding salt in a repeating cell unit that consists of three individual cells containing the salt solution, the acid and the base, and three membranes, that is, a cation-exchange, an anion-exchange, and a bipolar membrane. In industrial-size stacks 50–100 repeating cell units may be placed between two electrodes. The key element in electrodialysis with bipolar membranes is the bipolar membrane. Its function is illustrated in Figure 5.11(a), which shows a bipolar membrane consisting of an anion- and a cation-exchange layer arranged in parallel between two electrodes. If a potential difference is established between the electrodes, all charged components will be removed from the interphase between the two ion-exchange layers. If only water is left in the solution between the membranes, further transport of electrical charges can be accomplished only by protons and hydroxyl ions that are in a bipolar membrane are regenerated due to the water dissociation taking place in a very thin, that is, 4–5-nm thick transition region between the cation- and anion-exchange layers as shown in Figure 5.11(b). The water dissociation equilibrium is given by: 2H2 O , H3 Oþ þ OH

j107

j 5 Fundamentals in Electromembrane Separation Processes

108

Figure 5.11 Schematic diagram illustrating the function of a bipolar membrane showing (a) a bipolar membrane and (b) the 4–5-nm thick transition region at the interphase of the two cationand anion-exchange layers.

The energy required for the water dissociation can be calculated from the Nernst equation for a concentration chain between solutions of different pH values. It is given by: DG ¼ FDj ¼ 2:3 RTDpH

ð5:32Þ

Here, DG is the Gibbs free energy, F is the Faraday constant, the R is the gas constant, T is the absolute temperature and DpH and Dj are the pH value and the voltage difference between the two solutions separated by the bipolar membrane. For 1 mol L1 acid and base solutions in the two phases separated by the bipolar membrane DG is 0.022 [kWh mol1] and Dj is 0.828 [V] at 25 C. The transport rate of H þ and OH ions from the transition region into the outer phases cannot exceed the rate of their generation. However, the generation rate of H þ and OH ions in a bipolar membrane is drastically increased compared to the rate obtained in water due to a catalytic reaction [26, 27]. Therefore, very high production rates of acids and bases can be achieved in bipolar membranes. 5.4.2.1 Electrodialysis with Bipolar Membrane System and Process Design The design of an electrodialysis process with bipolar membranes is closely related to that of a conventional electrodialysis desalination process. Stack design in bipolar membrane electrodialysis The key component is the stack which in general has a sheet-ﬂow spacer arrangement. The main difference between an electrodialysis desalination stack and a stack with bipolar membranes used for the production of acids and bases is the manifold for the distribution of the different ﬂow streams. As indicated in the schematic diagram in Figure 5.10 a repeating cell unit in a stack with bipolar membranes is composed of a bipolar membrane and a cation- and an anion-exchange membrane and three ﬂow streams in between, that is, a salt

5.4 The Principle of Electromembrane Processes

Figure 5.12 Schematic diagram indicating the production of acids and bases from the corresponding salt in a stack with feed and bleed operation - See below for new figure.

solution, a base and an acid ﬂow stream. Since in most practical applications high acid and base concentrations are requested the stack is usually operated in a feed and bleed concept as shown Figure 5.12 Because of the relatively high concentrations of the acid and base as well as the salt solution the limiting current density is in general no problem and a bipolar membrane stack can generally be operated at very high current densities compared to an electrodialysis stack operated in desalination. However, membrane scaling due to precipitation of multivalent ions such as calcium or heavy-metal ions is a severe problem in the base-containing ﬂow stream and must be removed from the feed stream prior to the electrodialysis process with a bipolar membrane. Problems in the practical application of bipolar membrane electrodialysis In addition to the precipitation of multivalent ions in the base containing ﬂow stream and the stability of the ions in strong acids and bases a serious problem is the contamination of the products by salt ions that permeate the bipolar membrane. In particular, when high concentrations of acids and bases are required the salt contamination is generally high [28] as illustrated in Figure 5.13 that illustrates the conversion of

Figure 5.13 (a) Schematic diagram illustrating the contamination of acids and bases by salt due to the incomplete permselectivity of the bipolar membrane for salt ions; (b) experimentally determined salt concentration as a function of the acid and base concentration.

j109

j 5 Fundamentals in Electromembrane Separation Processes

110

Na2SO4 into H2SO4 and NaOH by electrodialytic water dissociation. Figure 5.13(a) shows the ion transport in the bipolar membranes. What is desired is a ﬂux of H þ and OH ions from the transition region due to the applied voltage into the outer phases. However, there is also an undesired transport of Na þ and SO42 ions through the bipolar membrane due to the incomplete permselectivities of the ion-exchange layers of the bipolar membrane. Since the permselectivity of the ion-exchange layers of the bipolar membrane decreases with increasing acid and base concentration due to the Donnan exclusion effect the contamination of the products is increasing with their concentration as demonstrated in Figure 5.13(b) that shows experimentally determined salt concentrations in the acid and base as a function of their concentration. The salt leakage through the bipolar membrane also effects the current utilization to some extent. However, the current utilization is mainly determined by the properties of the anion-exchange membrane, which has very poor retention for protons due to the tunneling mechanism of the proton transport as illustrated in the schematic drawing of Figure 5.14(a) that shows the undesirable transport of protons through the anionexchange membrane. The same is true for the hydroxide ions that can permeate the cation-exchange membrane. The net result is that H þ and OH ions generated in the bipolar membrane neutralize each other and thus reduce the current utilization. The ﬂuxes of the protons and hydroxide ions depend on their concentration. At high acid and bases concentrations the current utilization can reach uneconomically low values of less than 30% as indicated in Figure 5.14(b) that shows experimentally determined current utilization as a function of the acid and base concentration. 5.4.2.2 Electrodialysis with Bipolar Membrane Process Costs The determination of the costs for the production of acids and bases from the corresponding salts follows the same general procedure as applied for the costs in electrodialysis desalination. The contributions to the overall costs are the investmentrelated cost and the operating costs.

Figure 5.14 (a) Schematic diagram illustrating the decrease of the current utilization during the acid and base production due to the poor acid blocking capability of the anion-exchange membrane; (b) experimentally determined current utilization as function of the acid and base concentration.

5.4 The Principle of Electromembrane Processes

Investment costs in electrodialysis with bipolar membranes Investment costs include nondepreciable items such as land and depreciable items such as the electrodialysis stacks, pumps, electrical equipment, and monitoring and control devices. The investment costs are determined mainly by the required membrane area for a certain plant capacity. The required membrane area for a given capacity plant can be calculated from the current density in a stack that is in electrodialysis with a bipolar membrane not limited by concentration-polarization effects. The required membrane area for a given plant capacity is given by:

A¼

Qp F C p ix

ð5:33Þ

Here, A is the required membrane area, i is the current density, Q p is the product volume ﬂow, F is the Faraday constant, x is the current utilization and Cp is the concentration of the product. The required membrane area A refers actually to a unit cell area that contains a bipolar membrane, and a cation- and an anion-exchange membrane. Since in strong acids and bases the useful life of the bipolar membrane as well as the anion-exchange membrane is rather limited, the stack-related investment costs are dominating the total investment costs. Operating cost in electrodialysis with bipolar membranes The operating costs in electrodialysis with bipolar membranes are strongly determined by the energy requirements that are composed of the energy required for the water dissociation in the bipolar membrane and the energy necessary to transfer the salt ions from the feed solution, and protons and hydroxide ions from the transition region of the bipolar membrane into the acid and base solutions. The energy consumption due to the pumping of the solutions through the stack can generally be neglected. The total energy for the production of an acid and a base from the corresponding salt is as in electrodialysis desalination that has been discussed earlier given the total current passing through the stack and the voltage drop across the stack. The total energy required in electrodialytic water dissociation in a practical process is given by the current passing through the stack multiplied with the total voltage drop encountered between the electrodes.

Epro ¼ IDU t

ð5:34Þ

Here, Epro is the energy consumed in a stack for the production of an acid and a base, I is the current passing through a stack or a series of stacks, DU is the voltage applied across the stack, that is, between the electrodes, and t is the time of operation. The current passing through the stack can be derived by rearranging Equation 5.33. It is: I ¼ Ai ¼

Qp F Cpin Cpout x

ð5:35Þ

Here, A is a cell unit area. i is current density, I is the current, Qp is the ﬂow rate of the product, Cp is the concentration of the product, F is the Faraday constant, x the

j111

j 5 Fundamentals in Electromembrane Separation Processes

112

current utilization and the superscripts in and out refer to the in- and outlet of the stack. The voltage drop across the stack is the result of the electrical resistance of the membranes, that is, that of the cation- and anion-exchange membranes and the bipolar membranes and the resistances of the acid, the base- and the salt-containing ﬂow streams in the stack. In addition to the voltage drop required to overcome the various electrical resistances of the stack additional voltage drop is required to provide the energy for the water dissociation which is given by Equation 5.32. Assuming that the three cells of a cell unit in the stack have the same geometry and ﬂow conditions the total energy consumption in an electrodialysis stack is given by: 1 0 B D Ncell Acell x 2:3 RT DpHC C þ r am þ r cm þ r bm þ Epro ¼ Ncell Acell B @X L C Qp Cpout Cpin F 2 A i i i

12 0 Qp F Cpout Cpin At @ Ncell Acell x

ð5:36Þ

Here, Epro is the energy for the production of a certain amount of acid and base, I is the current passing through the stack, Ncell is the number of cell units in a stack, Acell is the are the concentration and the average concentration in a cell, D is cell unit area, C and C the thickness of the individual cells, and L is the equivalent conductivity, r is the area resistance, x is the current utilization, R is the gas constant, T the absolute temperature, F the Faraday constant, and DpH is the difference in the pH value between the acid and base, the subscript p refers to product and the subscript i refers to salt, acid and base, The superscripts am, cm, and bm refer to the cation-exchange, the anionexchange, and the bipolar membrane, the superscript out and in refer to cell outlet and inlet, Q is the total ﬂow of the acid or base through the stack and t is the time. The term Q p FðCpout Cpin Þ=Acell Nst x is identical to the current density. This means that for a given stack design the acid and base production energy Epro is proportional to the i2. The average concentrations of the acid, the base, and the salt in the bulk solutions are the integral average of the solutions given by: out Ci ln X Ciin i ¼ ð5:37Þ C Ciout Ciin i The total costs of the electrodialytic water dissociation with bipolar membranes are the sum of ﬁxed charges associated with the amortization of the plant investment costs and of the operating costs which include energy and maintenance costs and all pre- and post-treatment procedures. The total costs are a function of the membrane properties, of the feed-solution composition, the required acid and base concentrations, and several process and equipment design parameters such as stack construction and operating current density.

5.4 The Principle of Electromembrane Processes

5.4.3 Continuous Electrodeionization

Continuous electrodeionization is widely used today for the preparation of highquality deionized water for the preparation of ultrapure water in the electronic industry or in analytical laboratories. The process is described in some detail in the patent literature and company brochures [29]. There are also some variations of the basic design as far as the distribution of the ion-exchange resin is concerned. In some cases the diluate cell is ﬁlled with a mixed bed ion-exchange resin, in other cases the cation- and anion-exchange resins are placed in series in the cell. More recently, bipolar membranes are also being used in the process. 5.4.3.1 System Components and Process Design Aspects The process design and the different hardware components needed in electrodeionization are very similar to those used in conventional electrodialysis. The main difference is the stack construction. In a continuous electrodeionization stack the diluate cells and sometimes also the concentrate cells are ﬁlled with an ion-exchange resin. The different concepts used for the distribution of the cat- and anions in the cell are illustrated in Figures 5.15(a) and (b). In the conventional electrodeionization process the diluate cell is ﬁlled with a mixed-bed ion-exchange resin with a ration of cation- to anion-exchange resin being close to 1 as shown in Figure 5.15(a). The mixed-bed ion-exchange resin in the diluate cells of the stack removes the ions of a feed solution. Due to an applied electrical ﬁeld the ions migrate through the ionexchange bed towards the adjacent concentrate cells and highly deionized water is obtained as a product. The ion-exchange resin increases the conductivity in the diluate cells to such an extent that the stack resistance is signiﬁcantly lower and the limiting current density higher than in a conventional electrodialysis stack. Compared to the deionization by a conventional mixed-bed ion-exchange resin the continuous electrodeionization has the advantage that no chemicals are needed for the regeneration of the ion-exchange resins, which is time consuming, labor intensive, and generates a salt-containing wastewater. But the continuous electrodeionization using a stack with mixed-bed ionexchange resins in the diluate has also disadvantages. The most important one is the poor removal of weak acids and bases such as boric or silicic acid [30]. Much better removal of weakly dissociated electrolytes can be obtained in a system in which the cation- and anion-exchange resins are placed in a stack in separate beds with a bipolar membrane placed in between, as illustrated in Figure 5.15(b), which shows a diluate cell ﬁlled with a cation-exchange resin facing towards the cathode separated by a bipolar membrane from a diluate cell facing the anode. A cationexchange membrane, a cation-exchange resin, a bipolar membrane, an anionexchange resin, an anion-exchange resin, and a concentrate cell form a repeating unit between two electrodes. The main difference between the electrodeionization system with the mixed-bed ion-exchange resins and the system with separate beds is that in mixed-bed electrodeionization systems anions and cations are simultaneously removed from the feed

j113

Figure 5.15 Schematic drawing illustrating different stack concepts used in continuous electrodeionization, (a) conventional stack with diluate cells filled with a mixed-bed ion-exchange resin, (b) stack with cation-exchange and anion-exchange resins in different diluate cells and regeneration of the ion-exchange resins by H þ and OH ions generated in a bipolar membrane.

5.4 The Principle of Electromembrane Processes

while the solution leaving the diluate cell is neutral. In the electrodeionization system with separate ion-exchange beds and bipolar membranes the cations will ﬁrst be exchanged by the protons generated in the bipolar membrane with the result that the solution leaving the cation-exchange bed is acidic. This solution is then passed through the cell with the anion-exchange resin where the anions are exchanged by the OH ions generated in the bipolar membrane and the solution is neutralized, and at the exit of the anion-exchange-ﬁlled cell the solution is also neutral. Both the mixed and the separate bed ion-exchange continuous electrodeionization systems are widely used today on a large industrial scale. 5.4.3.2 Operational Problems in Practical Application of Electrodeionization In addition to the problems of removing weak acids or bases in the electrodeionization system the mixed bed ion-exchange resin there are problems of uneven ﬂow distribution in the ion-exchange resin beds that lead to poor utilization of the ionexchange resins. The fouling of the ion-exchange resins by organic components such as humic acids, and bacterial growth on the surface of the resin is a problem that requires a very thorough pretreatment of the feed solution to guarantee a longterm stability of the system. The effect of the cell geometry, that is, the ratio of its length to width and thickness has been studied extensively and is described in various patents. 5.4.4 Other Electromembrane Separation Processes

In addition to the processes discussed so far there are two more electromembrane separation processes in which the driving force is not an externally applied electrical potential but a concentration gradient. The processes are referred to as diffusion dialysis and Donnan dialysis. Diffusion dialysis is utilizing anion- or cation-exchange membranes only to separate acids and bases from mixtures with salts. Donnan dialysis can be used to exchange ions between to solutions separated by an ionexchange membrane. Both processes have so far gained only limited practical relevance [4] and will not be discussed in this chapter.

List of Symbols

Roman Letters

a a A C C D

constant [] activity [mol m3] area [m2] concentration [mol m3] average concentration [mol m3] diffusion coefﬁcient [m2 s1]

j115

j 5 Fundamentals in Electromembrane Separation Processes

116

E F G i I ilim J Je k L l n N p q Q r R R R S T t T U u V V z z Zb

energy [A V s] Faraday constant [A s eq1] Gibbs free energy [J] current density [A m2] current [A] limiting current density [A m2] ﬂux [mol m2 s1] ﬂux of electrical charges [A m2] coefﬁcient [various] coefﬁcient [mol2 N1 m2 s1] length [m] number [] number [] pressure [Pa] area [m2] volume ﬂow rate [m3 s1] area resistance [W m2] electrical resistance [W] average electrical resistance [W] gas constant [J mol1 K1] conductivity [W1] temperature [K] time [s] transport number [] electrical potential [V] ion mobility [m2 s1 V1] partial molar volume [m3 mol1] volume [m3] directional coordinate [m] charge number [eq mol1] boundary layer thickness [m]

Greek Letters

Y D D ~ m j k Leq Lmol m n

membrane permselectivity [] cell thicknes [m] difference [] electrochemical potential [A V s mol1] electrical potential [V] speciﬁc conductivity [W1 m1] equivalent conductivity [m2 W1 eq1] molar conductivity [m2 W-1 mol-1] chemical potential [J mol1] stoichiometric coefﬁcient []

5.4 The Principle of Electromembrane Processes

r x

speciﬁc resistance [W m] current utilization []

Subscripts

a c cell co eff des e ﬁx i k m am cm p pro s spec st tot w

anion cation cell or cell pair co ion efﬁciency desalination electric charge ﬁxed ion component component membrane anion-exchange membrane cation-exchange membrane pumping product salt speciﬁc stack total water

Superscripts

am b b bm c cm e d f fc fd in m out p s

anion-exchange membrane bulk solution constant bipolar membrane concentrate cation-exchange membrane electrode rinse diluate feed feed concentrate feed diluate inlet membrane outlet product solution

j117

j 5 Fundamentals in Electromembrane Separation Processes

118

References 1 Spiegler, K.S. (1956) Electrochemical operations, in Ion-Exchange Technology (eds F.C. Nachod and J. Schubert), Academic Press, New York. 2 Wilson, J.R. (ed.) (1960) Demineralization by Electrodialysis, Butterworth Scientiﬁc Publications, London. 3 Schaffer, L.H. and Mintz, M.S. (1966) Electrodialysis, in Principles of Desalination (ed. K.S. Spiegler), Academic Press, New York, pp. 3–20. 4 Strathmann, H. (2004) Ion-Exchange Membrane Separation Processes, Elsevier, Amsterdam, The Netherlands. 5 Liu, K.J., Chlanda, F.P. and Nagasubramanian, K.J. (1977) Use of bipolar membranes for generation of acid and base: An engineering and economic analysis. Journal of Membrane Science, 2, 109. 6 Eisenberg, A. and Yeager, H.L. (1982) Perﬂuorinated Ionomer Membranes, ACS Symposium Series 180, American Chemical Society, Washington, DC. 7 Flett, D.S. (1983) Ion-Exchange Membranes, E. Horwood Ltd, Chichester, UK. 8 Bergsma, F. and Kruissink, Ch.A. (1961) Ion-exchange membranes. Fortschritte der Hochpolymer Forschung, 21, 307. 9 Helfferich, F. (1962) Ion-Exchange, McGraw-Hill, London. 10 Sata, T. (1986) Recent trends in ionexchange research. Pure and Applied Chemistry, 58, 1613. 11 Grot, W. (1975) Perﬂuorinated cation exchange polymers. Chemie Ingenieur Technik, 47, 617. 12 Zschocke, P. and Quellmalz, D. (1985) Novel ion exchange membranes based on an aromatic polyethersulfone. Journal of Membrane Science, 22, 325. 13 Simons, R. (1993) Preparation of a high performance bipolar membrane. Journal of Membrane Science, 78, 13. 14 Wilhelm, F.W. (2000) Bipolar membrane preparation, in Bipolar Membrane Technology (ed. A.J.B. Kemperman),

15

16

17

18 19

20

21

22 23

24

25

26

Twente University Press, Enschede, The Netherlands. Pourcelly, G. (2002) Conductivity and selectivity of ion exchange membranes: structure-correlations. Desalination, 147, 359. Spiegler, K.S. (1958) Transport processes in ionic membranes. Transactions of the Faraday Society, 54, 1408. Kedem, O. and Katchalsky, A. (1961) A physical interpretation of the phenomenological coefﬁcients of membrane permeability. The Journal of General Physiology, 45, 143. Planck, M. (1890) Ann Physik u Chem, NF, 39, 161. Strathmann, H. (1995) Electrodialysis and related processes, in Membrane Separation Technology (eds R.D. Nobel and S.A. Stern), Elsevier, Amsterdam, pp. 213–281. Schaffer, L.H. and Mintz, M.S. (1966) Electrodialysis, in Principles of Desalination (ed. K.S. Spiegler), Academic Press, New York, pp. 3–20. Huffmann, E.L. and Lacey, R.E. (1972) Engineering and economic considerations in electromembrane processing, in Industrial Processing with Membranes (eds R.E. Lacey and S. Loeb), John Wiley & Sons, New York, pp. 39–55. Katz, W.E. (1979) The electrodialysis reversal (EDR) process.Desalination, 28, 31. Lee, H.J., Safert, F., Strathmann, H. and Moon, S.H. (2002) Designing of an Electrodialysis desalination plant. Desalination, 142, 267. Mani, K.N. (1991) Electrodialysis water splitting technology. Journal of Membrane Science, 58, 117. Kemperman, A.J.B. (ed.) (2000) Handbook on Bipolar Membrane Technology, Twente University Press, Endschede, The Netherlands. Strathmann, H., Krol, J.J., Rapp, H.J. and Eigenberger, G. (1997) Limiting current density and water dissociation in bipolar

References membranes. Journal of Membrane Science, 125, 123. 27 Mafe, S., Ramirez, P., Alcaraz, A. and Aguilella, V. (2000) Ion transport and water splitting in bipolar membranes, in Handbook Bipolar Membrane Technology (ed. A.J.B. Kemperman), Twente University Press, Endschede, The Netherlands. 28 Rapp, H.J. (1965) Die Elektrodialyse mit bipolaren Membranen – Theorie und Anwendung, PhD-Thesis, University of Stuttgart, Germany.

29 Ganzi, G.C. (1988) Electrodeionization for high purity water production, in New Membrane Materials and Processes for Separation, 84 (eds K.K. Sirkar and D.R. Lloyd), AIChE Symposium Series, p. 73, American Chemical Society, Washington, DC, USA. 30 Grabowskij, A., Zhang, G., Strathmann, H. and Eigenberger, G. (2006) The production of high purity water by continuous electrodeionization with bipolar membranes. Journal of Membrane Science, 281, 297.

j119

j121

6 Fouling in Membrane Processes Anthony G. Fane, Tzyy H. Chong, and Pierre Le-Clech

6.1 Introduction

This chapter reviews membrane fouling with particular reference to the pressuredriven liquid-phase membrane processes where the solvent is water. The processes of interest are low-pressure microﬁltration (MF) and ultraﬁltration (UF) and highpressure nanoﬁltration (NF) and reverse osmosis (RO). Fouling presents as a decrease in membrane performance with a loss in solvent permeability and changes to solute transmission. Fouling is caused by deposition of feed components, or growth (as in biofouling and scale formation) onto or into the membrane; it is a widespread and costly problem. The foulant–membrane interaction depends on the nature of the foulant, the membrane and the operating environment. This section provides an overview of fouling and describes various generic fouling mechanisms. It compares the fouling proﬁles of constant pressure vs constant ﬂux and crossﬂow vs. deadend. The concepts of critical and sustainable ﬂux are introduced. Fouling aspects of low-pressure microporous membranes and high-pressure nonporous membranes are described in Sections 6.2 and 6.3, respectively. 6.1.1 Characteristics of Fouling

The basic relationship between ﬂux and driving force is given in Equation 6.1 (Table 6.1). When fouling occurs an additional resistance, RF, is imposed and in some cases (with NF and RO) it may increase DP in Equation 6.1 (see Sections 6.1.3 and 6.3.4). Increasing RF and/or DP causes a ﬂux decline at constant DP (transmembrane pressure, TMP) (Figure 6.1(a)) or causes TMP to rise at constant ﬂux (Figure 6.1(b)). The ﬂux–time proﬁle (Figure 6.1(a)) can be misleading. For membrane M1, with initial ﬂux Ji an increment of resistance DRF reduces ﬂux to 0.5 Ji, then for similar DRF to 0.33 Ji, 0.25 Ji, and so on so ﬂux decline appears to be slowing down. A clearer picture emerges by calculating the changes in DRF. Figure 6.1(a) also

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 6 Fouling in Membrane Processes

122

Table 6.1 Fouling and polarization relationships

Description

Relationship

DPDP mðRm þ RF Þ

Equation

Flux-driving force resistances

J¼

Membrane resistance

Rm ¼ f fNpore ; dpore ; em ; Lpore g

(6.2)

Cake resistance

Rc ¼ mac

(6.3)

Speciﬁc cake resistance

ac ¼

CP ﬁlm model (for condition of complete retention of solute)

180ð1ec Þ ðrp d2p e3c Þ

Cw J ¼ k ln Cb Cw J ¼ exp Cb k

(6.1)

(6.4)

(6.5)

(6.6)

illustrates that the low ﬂux membrane M2, seems to decline less dramatically for the same DRF. An intrinsically highﬂux membrane appears more sensitive to fouling than a low ﬂux membrane. To make a fair comparison the membranes need to be tested at similar ﬂuxes (see Section 6.1.4) and fouling quantiﬁed as changes in DRF or TMP (at ﬁxed ﬂux). Figure 6.1(b) shows a steady fouling (i) and an example with a sudden TMP jump (ii) , which is characteristic of MBRs (see Section 6.2.4.2). The other detrimental effect of fouling is that it changes the separation properties of the membrane. For microporous membranes that transmit some species and retain others the effect of fouling is usually to increase retention of partially transmitted species. This is because fouling leads to pore closure or blockage (see Section 6.1.3), making the membrane tighter. In some cases, such as water

Figure 6.1 Fouling profiles. (a) Constant pressure (b) Constant flux.

6.1 Introduction

treatment and membrane bioreactors, this can be beneﬁcial, giving a greater removal of virus or organic species. In other cases, such as the food and biotechnology industries, this tightening can be a problem if membrane fractionation of species is desired. For nonporous membranes the fouling is a surface layer or cake (see Section 6.1.3) that increases polarization of retained and partially retained species at the membrane surface. The partially retained species then tend to show a higher concentration in the permeate, so observed retention decreases due to fouling. This is opposite to the trend for microporous membranes. 6.1.2 Causes of Fouling

Most dissolved or suspended species have the potential to foul membranes. Fouling interactions could be physicochemical adsorption from solution, precipitation of sparingly soluble salts, growth of bioﬁlms and deposition of suspended matter onto and into the membrane. Table 6.2 lists generic foulants and foulant control by pretreatment or feed-adjustment strategies. Examples of the various forms of fouling are given in Sections 6.2 and 6.3. The factors that encourage or exacerbate fouling are inadequate pretreatment, inadequate ﬂuid management (hydrodynamic environment), excessive ﬂux and unsuitable membrane properties. Pretreatment. Most feed streams are mixtures with varying characteristics. In many cases there are foulants present that can be minimized by pretreatment. For example, seawater fed to RO desalination plant usually contains turbidity and micro-organisms, which can be partially mitigated by preﬁltration (media or membranes) and Table 6.2 Foulants and foulant control.

Generic foulant

Type of fouling

Membrane

Control of foulant

Inorg. Ions Insoluble salts (Ca etc.) Inorg. Ions Calcium and so on Organics NOM, humics

Scale

NF RO

Concentration, pH, antiscalant

NOM binding

MF/UF NF

Concentration, pH, coagulants

Cake & bioﬁlms

NF, RO

NOM, humics Protein (food) Protein & polysaccharide Particulates Colloids(<1 mm)

Cake & pore fouling Cake & pore fouling Cake & pore fouling

MF, UF UF MF, UF MBR

Adsorb, biotreatment, MF/UF þ coagulants Coagulant pH MBR bioprocess control

Cake

NF, RO

Colloids(<1 mm) Biological solids

Cake & pore fouling Bioﬁlm

MF, UF NF, RO

Coagulant þ media ﬁlter or þ MF/UF Coagulant Chlorination, nutrient removal, coagulant þ MF/UF

j123

j 6 Fouling in Membrane Processes

124

chlorination (followed by dechlorination) respectively. If scale formers are present they can be suppressed by addition of antiscalants. MBR membranes handling municipal wastewater are susceptible to damage and this can be avoided by ﬁnescreen pretreatment. In some cases, such as water treatment, the foulants can be controlled by chemical pretreatment so that ﬁltration is of large ﬂoc, rather than macrosolutes or colloids. Where pretreatment is inadequate fouling usually results and leads to frequent cleaning or system upgrade. Fluid Management.Concentration polarization (CP) at the membrane surface is a result of the separation of feed and permeate (mainly water) at the interface. Fouling is a consequence of CP. The concentration of species at the surface (Cw) depends on the imposed ﬂux (J) and the boundary layer mass-transfer coefﬁcient (k) as indicated by the CP ﬁlm model (Equations 6.5 and 6.6 in Table 6.1). Tangential crossﬂow is used to limit CP, and does so by increasing k. If ﬂuid management is inadequate for the feed serious CP can lead to fouling. The reasons for inadequate ﬂuid management include badly selected operating conditions, loss of feed ﬂow due to permeate removal or ﬂow maldistribution due to poorly designed membrane module or blockages. In deadend ﬂow (Section 6.1.5), which is applied to dilute feeds, there is no crossﬂow but it is still important to ensure homogeneous ﬂow distribution and ﬂux to avoid localized fouling. Flux. The ﬁlm model (Equation 6.6) illustrates that increasing ﬂux has an exponential effect on CP. If we accept that fouling is a consequence of CP the impact of excessive ﬂux is obvious. As a result high ﬂux membranes tend to be short lived and foul unless improved ﬂuid management is able to enhance k. Selection of the appropriate ﬂux and crossﬂow velocity is a trade-off between capital and operating costs (see cost of fouling below). Membranes. The interactions between the membrane and the potential foulants inﬂuence the degree of fouling. For example, if the pore size of a microﬁlter and particulate foulants overlap the particles can enter the membrane and block pores (see Section 6.1.3), and this can cause irreversible fouling. If the surface charge of the membrane and the dominant feed species are opposite it is likely that fouling will occur. Similarly, hydrophobic membranes are prone to fouling by hydrophobic components, including proteins and lipids. A useful rule of thumb is that membranes that are smooth, hydrophilic, of low net charge and narrow pore-size distribution are less susceptible to fouling. However, there are many exceptions to this rule and membrane selection may involve comparison trials. The cost of fouling has at least 4 components, represented by, cFouling ¼ cCleaning þ cPower extra þ cProduction loss þ cMembrane replace

ð6:7Þ

cCleaning includes cost of chemicals, disposal and labor. cPower extra represents the effect of increased operating pressure. cProduction loss includes the effect of ﬂux decline on throughput as well as down time for cleaning. In respect of cMembrane replace it should be noted that cleaning events may be the harshest environment for the membrane and replacement tends to increase with cleaning frequency. Fouling costs are application speciﬁc but could range from a few % to 10s of % of processing costs.

6.1 Introduction

Figure 6.2 Membrane-fouling mechanisms.

6.1.3 Fouling Mechanisms and Theory

Figure 6.2 depicts 3 ways in which microporous membranes can foul; (a) pores can suffer closure or restriction, (b) pores or porosity can be blocked or plugged and (c) a surface cake or layer can cover the membrane. All three mechanisms could apply, probably in sequence (a) then (b) followed by (c). Nonporous membranes are fouled by cake or surface layers (c). It may be possible to do a membrane autopsy to identify the foulant(s) and fouling mechanism. For microporous membranes the blocking law analysis [1], which uses permeate volume (V) vs. time (t) data, can supplement the observations. The generalized relationship at constant pressure and in dead-end ﬁltration mode gives, n d2 t dt ¼ K ð6:8Þ dV 2 dV where n ¼ 0 for cake ﬁltration, n ¼ 1.5 for pore closure (standard blocking) and n ¼ 2 for pore plugging (complete blocking). Linearized forms of the blocking laws have been used to investigate fouling [2]. Some useful relationships describing fouling are given in Table 6.1 with the basic relationship in Equation 6.1. Fouling can increase membrane resistance Rm (Equation 6.2), particularly for microporous membranes where loss of pores (Npore) due to plugging and reduction in pore size (dpore) due to restriction may cause irreversible fouling. If fouling is irreversible it implies a resistance not readily removed by cleaning. Fouling due to cake formation adds resistance, RF, (Equation 6.3) and this depends on the foulant load (m) and speciﬁc cake resistance (ac) that is increased as particle size (dp) and cake voidage (ec) decrease (Equation 6.4). Fouling can also increase DP and diminish the driving force. This is known as cake-enhanced osmotic pressure (CEOP) and occurs when retained solutes, such as salts in RO, have hindered backdiffusion in the cake layer so that CP increases and local osmotic pressure rises. CEOP is a feature of fouling in NF and RO and is discussed in Section 6.3. 6.1.4 Critical and Sustainable Flux

The concept of critical ﬂux ( JCRIT) was introduced by Field et al. [3] and is based on the notion that foulants experience convection and back-transport mechanisms and that there is a ﬂux below which the net transport to the membrane, and the fouling, is negligible. As the back transport depends on particle size and crossﬂow conditions the JCRIT is species and operation dependent. It is a useful concept as it highlights the

j125

j 6 Fouling in Membrane Processes

126

role of ﬂux as a driver of fouling. However, in many practical cases JCRIT is very low or difﬁcult to identify. The sustainable ﬂux is an alternative concept and refers to the ﬂux below which the rate of fouling is economically acceptable. A recent review [4] examines these two concepts. 6.1.5 Fouling and Operating Mode

The pressure-driven membrane processes can be operated at ﬁxed pressure (FP) or ﬁxed ﬂux (FF), and FP tends to be lab and small scale and FF is large-scale commercial. Fouling for FP shows as a ﬂux decline and for FF as TMP rise (Figure 6.1(b)). The fouling kinetics differ since FP becomes self-limiting as ﬂux-driven fouling slows down, whereas for FF it is self-accelerating as foulants steadily accumulate and concentration polarization accelerates. These differences mean that extrapolation of FP trends to FF requires caution. As noted in Section 6.1.2, in most applications the control of CP, and fouling, dictates the use of crossﬂow. However, for dilute feeds and low-pressure membranes it has been accepted that batch cycles of deadend operation with solids accumulation removed by periodic backwash requires potentially lower energy. Usually, deadend is at FF and the TMP cycles from a minimum to maximum or over a speciﬁed cycle time during the batch. If fouling occurs it is evident through a steady rise in TMPmin or Rm. Occasional chemical cleaning may restore Rm.

6.2 Low-Pressure Processes

In the context of this review low-pressure membranes include the microporous MF and UF processes. Applications of MF include beverage clariﬁcation, cell harvesting, wastewater treatment by membrane bioreactors (MBRs), water treatment and pretreatment prior to RO. UF covers similar applications plus protein concentration (food and dairy) and other macrosolute separations. Crossﬂow, supplied by pumping, stirring or two-phase ﬂow, is generally applied, except when very low solids content permits deadend ﬁltration with frequent backwash, as in water treatment, RO pretreatment and clariﬁcation processes. The liquid streams to be treated by membrane processes usually contain a complex mixture of particulate, colloidal and soluble materials. Before considering two complex feeds (activated sludge and surface water) in Section 6.2.4, general trends observed during the ﬁltration of speciﬁc compounds are discussed. 6.2.1 Particulate Fouling

Examples of particles to be ﬁltered through low-pressure membranes include casein micelles from milk, lattices from paint, biomass ﬂocs from activated sludge, bacteria

6.2 Low-Pressure Processes

from cell broth and suspended solids from water or wastewater. In the early stages of the ﬁltration, the particles with sizes much smaller than the pores can enter and block the pores (Figure 6.2). Particles larger than the pores of the membrane are totally rejected and may be partially transported back to the bulk. However, the formation of a cake layer on the membrane surface gradually occurs as the ﬁltration proceeds, provided the local ﬂux exceeds the critical ﬂux of the particle (see Section 6.1.4). A high concentration of solids in the feed to be treated can be responsible for an increase in viscosity and a decrease in the translational diffusion coefﬁcient (and the resulting Brownian backdiffusion), both leading to higher concentration polarization and fouling propensity [5]. Brownian diffusion is more applicable to ﬁne colloids (Section 6.2.2) and decreases with particle size, being low for particles larger than about 0.5 mm. However other mechanisms have been proposed for the back transport of larger particles in crossﬂow microporous membrane processes for both laminar and turbulent conditions. These mechanisms include shear-induced diffusion (migration of interacting retained particles in the direction of decreasing particle concentration), inertial lift (motion of particles across a nonuniform shear ﬁeld to an equilibrium position away from the channel wall) and ﬂowing cake [5]. The key aspect of these particulate back-transport models is the prediction of more signiﬁcant fouling in the cases of small particles and at lower crossﬂow velocities. Because of their generally reversible nature, fouling by large particles is usually efﬁciently removed by physical methods such as membrane relaxation (ﬁltration is paused) and backwashing (permeate is pumped in the reverse direction through the membrane). Furthermore, many studies have reported the efﬁcient and optimized use of aeration for fouling limitation in submerged low-pressure membrane processes. Finally, improved pretreatment of the feed by coarse ﬁltration or by coagulation to increase particle size can also be considered to reduce the fouling potential of the particle-based compounds. 6.2.2 Colloidal and Macrosolute Fouling

There is a general consensus that colloidal solutions are composed of small particles whose size could range from 1 nm to 1 mm. Colloids experience doublelayer interactions. For example, when silica particles (0.14 mm) were used as model colloids, the ionic strength of the feed strongly inﬂuenced the fouling characteristics by increasing the cake packing density, leading to a lower efﬁciency of the backwash process used to remove the fouling layer [6]. McDonogh et al. [7] showed experimentally and by modeling that the limiting ﬂuxes and cake resistances of colloidal suspensions in UF varied signiﬁcantly with ionic environment as this alters particle–particle interaction through the zeta potential. More discussion of colloidal species is given in Section 6.3.1. With the majority of MF/UF processes based on the ﬁltration of aqueous solutions, the unavoidable presence of dissolved solutes (salts, organic macromolecules of various nature and sizes) in the feed strongly affects the membrane performances. They can be directly responsible for membrane fouling through mechanisms like

j127

j 6 Fouling in Membrane Processes

128

adsorption of macromolecular substances or precipitation and deposition of inorganic salts on the membrane surface [5]. Due to their changing nature and large diversity, the naturally occurring polymers (e.g., proteins, carbohydrates and humic substances) present signiﬁcant complexity to the fouling mechanisms (see case studies in Section 6.2.4), and their ﬁltration through MF/UF have been extensively studied in the literature [8, 9]. Interaction with ions, such as calcium, can also exacerbate the fouling by these macrosolutes. Overall, when considering the effect of dissolved material on membrane fouling, the osmotic pressure of the dissolved compounds has to be taken into consideration, along with the interactions they may have with the particles and colloids in the suspension. 6.2.3 Biofouling and Biofilms

During the early stages of ﬁltration of biomass suspensions, attachment of soluble microbial products (SMP) on the membrane surface through adhesive forces can be observed. This phenomenon has also been reported during passive adsorption when no ﬁltration occurs. The initial conditioning ﬁlm participates in the reduction of the hydraulic performance, and facilitates further cohesive attachment of colloids and/or particles (including bacteria) on the SMP-covered membrane. The newly immobilized bacteria are then able to grow and colonize the membrane surface, forming a bioﬁlm layer [10]. As permeation continues, the soluble compounds in the feed are transferred through the bioﬁlm structure, providing a constant supply of nutrients and dissolved oxygen (when available) to the growing bioﬁlm. The bioﬁlm formation on the membrane surface highlights the signiﬁcant role played by the strong interactions existing between particulates and macromolecular components, which can be observed in any multicomponent feed. Biofouling in RO is discussed in Section 6.3.2. 6.2.4 Case Studies 6.2.4.1 Water Treatment and Membrane Pretreatment As more potable-water treatment plants rely on NF/RO technologies, pretreatment systems able to efﬁciently and reliably remove suspended solids from the feed water are a critical feature of the overall plant design. A similar requirement pertains to water reclamation plant (used water to high-quality/indirect potable water). Conventional pretreatment processes include coagulation, followed by sedimentation and/or sand ﬁltration, and more recently membrane technology like MF/UF. Indeed, the number of plants based on the use of MF/UF for this application tends to rise as lowpressure membranes remain the only technology able to remove pathogens and bacteria (and decrease biofouling downstream) more effectively than other conventional processes without the need for chemical addition [11]. These applications of low-pressure membranes are characterized by low solids feeds and the trend to use deadend ﬁltration with backwash (Section 6.1.5).

6.2 Low-Pressure Processes

The presence of NOM in surface water and its effect on fouling of low-pressure membranes has been discussed in numerous publications (see recent review [12]). The complex and labile nature of these compounds remains a challenge for researchers to fully understand their fouling mechanisms. As expected, most of the parameters deﬁning the ﬁltration process (membrane and feed characteristics and operating conditions) can inﬂuence the propensity of NOM fouling. Within this long list, ionic strength, pH of the feed and presence of divalent cations, such as calcium, have been listed as the main factors affecting the degree of NOM fouling. Due to the wide size distribution of NOM species, fouling mechanisms include membrane adsorption, leading to pore closure and restriction, and cake formation. For water-treatment applications, MF or UF membranes are conﬁgured in submerged or contained (pressurized) systems and usually operated in deadend mode. Cleaning involves hydraulic backwashing (for example, 1 min every 30 min [11] often accompanied by air scouring. During the ﬁltration of a water source containing traces of bacteria, a direct relationship between EPS levels in the feed and the required frequency of backwashing was observed [13]. This study also highlighted that the characteristics of the colloidal fouling were dependent on the value of the applied pressure, which determined maximum ﬂux, rather than the mode of deposition (i.e., constant ﬂux or constant TMP operation). Through advanced characterization of the temporal changes in the fouling nature observed with submerged MF used for surface water, Yamamura and coworkers [11] proposed a detailed fouling mechanism where the membrane pores are ﬁrst covered and narrowed by large biopolymer species, followed by humic substances and divalent cations that further block the narrowed pores. After that a cake layer builds up on the membrane surface. The size of the particles to be ﬁltered plays a signiﬁcant role in the type of fouling obtained in the deadend mode: small particles at around 0.1 mm creating a more compact cake with higher speciﬁc resistance (Table 6.1, Equation 6.4) compared to those obtained with larger particles up to 1 mm [14]. As for any membrane process, the applied ﬂux has a crucial role, and has been found to speciﬁcally determine the backwashing frequency in dilute feed deadend MF applications. Based on ﬁltration theory the cycle time from low to high TMP is proportional to (1/ﬂux)2 [15]. Since energy use depends on backwash frequency and membrane area depends on imposed ﬂux, there is a trade-off between operating and capital costs and an optimal ﬂux can be obtained [15]. 6.2.4.2 Membrane Bioreactor (MBR) The idea of placing a microporous membrane in direct contact with activated sludge may have been considered pioneering and risky 40 years ago, but the recent widespread application of the MBR has since been proven sustainable, when proper operating conditions are applied. The complex and labile nature of the microbial population present in the MBR process presents new challenges for membrane operators. Activated sludge is composed of suspended solids (large biological ﬂocs, individual micro-organisms and inert particles), colloids and soluble materials (dissolved matter from the wastewater and soluble microbial products (SMP) excreted from biomass activities). A comprehensive review of MBR fouling is available [16].

j129

j 6 Fouling in Membrane Processes

130

Over recent years, the methods used to characterize the biomass have been diversiﬁed, allowing an improved understanding of the interaction between the compounds present in the activated sludge and the membrane. However, recent efforts to relate the EPS and SMP fractions (generally given in terms of protein and carbohydrate) to the MBR fouling propensity have not been universally successful in explaining the fouling mechanisms in MBRs [16]. Fouling mechanisms in the MBR are likely to include all 3 mechanisms in Figure 6.2 as well as biofouling. A 3 stage history for TMP rise, involving initial membrane fouling by adsorption and pore closure followed by a period of slow (sustainable) rise and ﬁnally a TMP jump (Figure 6.1(b)), has been described [17]. Many membrane suppliers have developed ﬁltration products speciﬁcally designed for MBR applications. Low-cost polymeric hydrophilic microporous membranes used in submerged conﬁgurations are generally suggested; their pore sizes range from 0.4 mm to 40 kDa [18]. While the large-pore MBRs rely on the formation of a fouling layer to produce high product quality, the intrinsic retention of UF-based systems are not ﬁltration-time dependent and show good performances from the early stage of the ﬁltration. The MBR operating conditions (SRT, HRT) are related to the quantity and quality of the wastewater to be treated and have a strong inﬂuence on the nature of the activated sludge [18]. Whilst their direct effect on the fouling propensity is still unclear there are indications that very short or very long SRT are more fouling prone. Air scour is universally used to control fouling in MBRs and the effect of air sparging on membranes has been reviewed in detail [19]. It has been frequently reported that once a certain air ﬂow rate is exceeded, no further signiﬁcant fouling limitation is observed and the rate of aeration is generally optimized. Membrane-cleaning strategies are numerous and generally remain proprietary information. Physical cleaning by relaxation or backwashing is used on a frequent basis but the efﬁciency tends to decrease with ﬁltration time. As irreversible fouling accumulates on the surface, chemical cleanings of various intensities (i.e., cleaner concentration used) can be applied on a weekly to yearly basis [20]. 6.3 High-Pressure Processes

The processes of interest are NFand RO where the membranes are either nanoporous or essentially nonporous. In these processes the fouling is a surface layer, the effects of which may be exacerbated by the high retention of solutes by the membrane. Operation is with crossﬂow and in industry ﬁxed ﬂux is commonly used. This section considers particulate fouling, biofouling and scale formation and then discusses the implications of cake enhanced concentration polarization on fouling outcomes. 6.3.1 Particulate and Colloidal Fouling

Particulate fouling in RO is most likely to be colloidal due to the formation of a colloidal deposit layer on the membrane surface. Examples of colloidal particles

6.3 High-Pressure Processes

include clay, iron oxide, silica, macromolecules (protein, polysaccharides and natural organic matters), and biocolloids (bacteria and viruses). These colloids are usually in the size range of 1–1000 nm (or 0.001–1 mm) and have an upper bound of 1–10 mm, but there is no strict margin [21]. In practice, the particulate fouling tendency of feed waters for RO is characterized by fouling indices, namely the silt density index (SDI) and the modiﬁed fouling index (MFI) [22, 23]. These methods are limited because they are based on the retention of colloids by MF/UF membranes under test conditions that differ from the actual RO process. Importantly these indices only predict the hydraulic resistance of the particulate cake layer, but do not measure the cake-enhanced osmotic pressure effect (refer to Section 6.3.4), which can be experienced in RO and NF. Colloidal fouling involves the transport of the foulant from the bulk ﬂuid to the membrane surface, followed by the particle-attachment process. In a crossﬂow RO system, particles are convected to and retained on the membrane surface due to the permeation ﬂux, J (perpendicular to the surface) while the crossﬂow (tangential to the surface) induces particles to be back transported from the membrane surface and into the bulk solution due to the concentration gradient. A particle-polarization layer is formed on the membrane surface that is similar to the concentrationpolarization (CP) layer of solute (dissolved ions) in RO. As noted above, the backtransport mechanisms in RO operation include Brownian diffusion (BD) and shear-induced diffusion (SID) [5]. Generally, for submicrometer size particles 2=3 (typically < 0.1 mm), BD is important and JBD / 1=ap whereas SID applies to 4=3 micrometer-sized particles (typically > 0.2 mm) and JSID / ap . When a particle is in the vicinity of the membrane/solution interface, colloid–membrane interactions could determine the attachment of the particle onto the membrane surface. The interactions may be attractive (e.g., van der Waals or hydrophobic attraction) or repulsive (e.g., electrical or steric repulsion), which are best described by the classical and extended Derjaguin–Landau–Verwey–Overbeek (DLVO) theory for colloid stability [21]. In general, the nature of the interaction is greatly dependent on the nature of the particle, membrane surface and solution chemistry such as particle size, zeta potential, ionic strength of the solution, and membrane surface roughness [24, 25]. Most RO applications are in a high or raised salinity environment (seawater, brackish water and reclaimed water) where the charge interactions are greatly suppressed. Therefore, the controlling phenomena in membrane fouling are mainly the particle transport step that is a function of convection (permeation drag) and the back-transport mechanisms BD and SID. The concept of critical ﬂux (Section 6.1.4) was introduced from observations in MF and UF [3], but intuitively it should also apply in a RO system. For colloidal fouling the onset and build up of a deposit layer is dependent on the net ﬂux, which is the difference between the operating ﬂux, J, and the critical ﬂux, Jcrit, or ( J Jcrit) [26]. In RO, a colloidal deposit on the membrane introduces an additional resistance, RF, and could also cause cake-enhanced concentration osmotic pressure (CEOP) [24]. The CEOP phenomenon is discussed in Section 6.3.4. Large-scale RO plants tend to be operated at a ﬁxed production rate, requiring a ﬁxed average ﬂux.

j131

j 6 Fouling in Membrane Processes

132

To compensate for fouling (RF and CEOP) it is necessary to increase transmembrane pressure (Figure 6.1(b)). The constant ﬂux strategy has important implications. Firstly, if J > Jcrit of foulant species, that species will continue to deposit. Secondly, as CP is (exponentially) ﬂux-driven (see Equation 6.6 in Table 6.1) it will rise due to CEOP in a self-accelerating fashion. This is in contrast to a ﬁxed-pressure strategy where the ﬂux declines, net convection of foulant drops and CEOP become self-limiting. 6.3.2 Biofouling

Biofouling in RO is a problem of formation of an unwanted bioﬁlm [27]. A bioﬁlm is deﬁned as, a surface accumulation, which is not necessarily uniform in time or space, which comprises cells immobilized at a substratum and frequently embedded in an organic polymer matrix of microbial origin [28]. Biofouling has long been recognized as one of the most problematic types of fouling in the RO process. Even after a 99.9–99.99% removal of bacteria by the use of microﬁltration as the pretreatment step, biofouling in RO cannot be eliminated as it only requires a few initial colonies on the membrane surface to eventually form a mature bioﬁlm [27]. In fact, it has been suggested that the majority of the bacterial population involved in bioﬁlm formation are viable but non culturable (VBNC) and are about 0.2 mm in size, making them difﬁcult to remove [29]. In order to survive, these recalcitrant bacteria adhere to the membrane surface, resuscitate (convert from a nonculturable state to a culturable form), multiply and secrete extracellular polymeric substances (EPS), and eventually form a mature bioﬁlm. EPS provides sorption sites for water, inorganic and organic solutes and particles and therefore could induce other types of fouling such as particulate fouling. In terms of bioﬁlm formation the major difference between RO and other systems is the presence of permeation ﬂux and the complex hydrodynamic conditions, for example, the use of feed spacers in the ﬂow channel. The challenge is to establish a relationship between bioﬁlm formation and ﬂux-mass transfer in a RO system. In general, the biofouling process can be divided into ﬁve stages: (1) the formation of a conditioning ﬁlm, (2) bacteria transport and adhesion, (3) bioﬁlm development and accumulation, and (4) bioﬁlm detachment. The ﬁrst step in bioﬁlm formation is the adsorption of macromolecules (e.g., humic substances, lipopolysaccharides, or other products of microbial turnover) onto the membrane surface, which is aided by the CP effect. The thin-ﬁlm organic layer is known as the conditioning ﬁlm. These adsorbed macromolecules can mask the original surface properties of the membrane and facilitate the attachment of micro-organisms onto the surface. Since most bacteria have an average size of about 1 mm, bacteria cell transport and adhesion is often treated equivalently to particle transport and attachment. Experimental studies that relate the initial stage of bacteria deposition onto the membrane surface with the transport models, for example, the shear-induced diffusion model, and the surface interaction forces, for example, DLVO and acid–base interactions have been demonstrated using direct observation methods [30, 31]. However, this may oversimplify

6.3 High-Pressure Processes

the transport and adhesion process as the actual bacteria cells are semisolid (some have an overall irregular shape due to the ﬂagella) and are surrounded by EPS. Once attached, bacteria cells may grow and proliferate into microcolonies, excreting EPS, colonizing free surface areas, and forming a mature bioﬁlm. Since the attached bacteria multiply at the expense of nutrients, so bioﬁlm growth could be signiﬁcantly accelerated at high CP level (under high J or low k conditions), which in turn controls the amount of nutrients at the membrane wall [32]. Detachment is an interfacial transfer process, which transfers cells from the bioﬁlm back to the bulk liquid. It was observed that bioﬁlm detachment increases with both ﬂuid shear stress and bioﬁlm mass [33]. Other factors such as the availability of nutrients can also determine the detachment of the bioﬁlm [34]. If the nutrients in the bioﬁlm are consumed, the situation will be unfavorable for the growth of micro-organisms. In natural-water systems, the bioﬁlm layer is limited by the balance between diffusion of nutrient and the rate of consumption of nutrient [35]. In RO, due to the presence of CP, the nutrient supply may be increased in the microenvironment near the membrane surface. The buildup of bioﬁlm on the membrane surface means an additional resistance to solvent ﬂow as well as the possibility of enhancement of CP level by the bioﬁlm, which is similar to the case of colloidal fouling [32, 36]. In general, the diffusivity is linked to the tortuosity factor of the bioﬁlm [37]. Hence, it is likely that the backdiffusion of solutes in the bioﬁlm on RO is hindered. The enhanced CP is important for two reasons. Firstly, the elevated concentration of solutes at the membrane wall means an increase in the osmotic pressure (CEOP) and hence a loss in the effective TMP. Secondly, the nutrient level is also enhanced and this will further accelerate the growth of the bioﬁlm [32, 36]. So, biofouling in RO becomes an interplay between CP and bioﬁlm development. 6.3.3 Scale Formation

Due to the retention of solutes by an RO membrane, the concentration of sparingly soluble mineral salts such as calcium carbonate, calcium sulfate, calcium phosphate, barium sulfate and so on can exceed the saturation level and cause scaling of the membrane surface. There are two pathways for membrane-scale formation, namely surface (heterogeneous) crystallization and bulk (homogeneous) crystallization [38]. In surface crystallization, the concentration of sparingly soluble salt at the membrane surface exceeds the solubility limit due to the CP effect. As a result, the membrane surface is blocked by the lateral growth of scale deposit, which greatly reduces the effective area for permeation [39]. Thus, for ﬁxed-ﬂux operation, the local ﬂux increases in order to achieve the same average ﬂux, so the local CP level is greatly enhanced due to the exponential relationship of ﬂux and CP (Equation 6.6 in Table 6.1). This means scaling can be exacerbated due to the higher degree of supersaturation. Alternatively, crystal particles form in the bulk phase through homogeneous crystallization when concentrations of salts surpass the saturation level due to high recovery of feed waters. The crystal particles are then transported by

j133

j 6 Fouling in Membrane Processes

134

Table 6.3 Crystallization and scaling relationships.

Description

Supersaturation ratio

Relationship

S¼

Equation

CW CS

(6.9)

s3n

Induction time for nuclei formation

logðtind Þ /

Rate of crystal growth

dmcry ¼ kcry ðCW CS Þn dt

T 3 ðlog SÞ2

! (6.10)

(6.11)

the permeation ﬂux and eventually form a porous cake layer on the membrane surface, which can be characterized by an increase in the hydraulic resistance [40], similar to particulate fouling. Furthermore, it should be noted that both scaling mechanisms could occur simultaneously in a RO system. This would be most likely if cake-enhanced concentration polarization was caused by a porous cake layer. CEOP effects could also contribute to fouling. Membrane-scaling phenomena can be well explained by the concept of crystallization; the relevant equations are given in Table 6.3. The supersaturation ratio is deﬁned in Equation 6.9, where Cw and Cs are the wall concentration (taking into account the effect of CP) and solubility limit of salt, respectively. So when S > 1, the salt has a potential to form a scale deposit. However, according to the classical nucleation theory, it takes time for the generation of nuclei, which is a precursor to crystal growth [41]. This time is called the induction time, tind, and it is related to the supersaturation ratio, as shown in Equation 6.10, where T is the temperature and sn is the surface energy. After the induction period, once stable nuclei have been formed, they begin to grow into crystals of ﬁnite size. The rate of crystal growth is commonly represented by Equation 6.11, where kcry and n are the rate constant and order of reaction, respectively. One of the strategies to control scaling is through control of the supersaturation level of the salts [42] and RO systems are operated at a recovery that limits the increase in the bulk concentration of salts. The concentration at the membrane wall can be manipulated by the ratio of J/k of the RO system (Equation 6.6). Other methods include, (1) the injection of acid into the feed stream to reduce the dissolved carbonate ions in order to control scaling due to calcium carbonate, (2) shifting the ion species to precipitate as a more soluble form of salts, for example, calcium forms more soluble complexes with EDTA than calcium carbonate, (3) altering the scale morphology by forming a rather thin deposit layer with less tenacity [43], which could have a lower hydraulic resistance, (4) retarding the crystallization process through extending the induction period of nucleation [44] and adding antiscalants.

6.3 High-Pressure Processes

Figure 6.3 (a) CP for clean membrane (b) CECP for fouled membrane.

6.3.4 Cake-Enhanced Osmotic Pressure

As mentioned earlier, the CEOP phenomenon arises due to cake-enhanced concentration polarization CECP. Backdiffusion of retained solutes is hindered because the solutes now need to diffuse through the tortuous paths within the cake layer. The solutes in this unstirred deposit layer are not exposed to crossﬂow and the concentration and osmotic pressure at the membrane surface are greatly enhanced. The concept is depicted in Figure 6.3 and the relevant equations are given in Table 6.4. Figure 6.3 also explains how fouling in RO (and NF) can decrease solute retention (see Section 6.1.1). The CECP phenomenon causes Cw to rise from Cw1 to Cw2 and this can raise permeate concentration Cp from Cp1 to Cp2. Thus fouling in RO reduces both water permeability and permeate quality. Equation 6.12 relates CP to ﬂux (J) and the effective mass-transfer coefﬁcient (keff ) for retained solute. For a clean membrane keff ¼ k, which is the boundary layer masstransfer coefﬁcient from Sherwood correlations [45, 46]. However, for a membrane Table 6.4 Cake-enhanced CP relationships.

Description

Concentration polarization

Relationship

CW J ¼ exp keff Cb

Equation

(6.12)

Effective mass transfer

1 1 1 ¼ þ keff k kc

(6.13)

Cake mass transfer

kc ¼

Dc dc

(6.14)

Diffusion in cake

Dc ¼

Cake tortuosity

tc ð1ln e2c Þ

Dec tc

(6.15)

(6.16)

j135

j 6 Fouling in Membrane Processes

136

with a cake layer the backdiffusion is hindered. Equation 6.13 shows how keff is related to the unfouled surface k and the mass transfer in the cake (kc). The cake mass transfer (Equation 6.14) is given by the diffusion in the cake, Dc, and the cake height, d c, where cake diffusion is the free solution diffusion modiﬁed by cake voidage, ec, and tortuosity, tc. An empirical relationship [47] relates tortuosity and voidage. For a typical value of voidage of 0.3 the value of Dc is only about 10% of D. For such a cake with d c ¼ 20 mm the kc value would be about 20% of the typical k values, so for fouling conditions k tends to be kc and keff ! kc. Thus, keff decreases as the fouling layer height dc grows, and hence CP increases. It should be noted that for a linear increase in fouling layer thickness, CP increases exponentially due to the form of Equation 6.12. For constant-ﬂux operation of large-scale RO the fouling continues and the required pump pressure (Equation 6.1, Table 6.1) will rise. In practice the pump delivery pressure will have an upper limit (DPmax) and once this is reached the ﬂux will not be sustainable. As ﬂux declines, net convection of foulant drops and concentration and CEOP become self-limiting. A pseudosteady state could be achieved with DP at DPmax and dc constant. However further ﬂux decline would tend to occur due to cake consolidation, accumulation of other foulants (NOM, ﬁne colloids) within the cake, bioﬁlm development or scale formation. The relative contribution to performance loss (TMP rise at constant ﬂux) of resistance RF or CEOP depends on the particle size. For particles >0.5 mm the resistance is relatively small (see effect of particle size in Equation 6.4) and CEOP due to cake height, dc, is the major effect [48]. It is also observed that bioﬁlms can contribute substantial CEOP effects as well as resistance, and in a recent biofouling study more than 50% of the required TMP rise was due to CEOP effects [32]. 6.4 Conclusions

Fouling is a major factor in the application of membranes. For both low-pressure and high-pressure membranes the degree of fouling is a complex function of feed characteristics, membrane properties and operating conditions. However, much is now known about fouling and how it can be controlled. The key to low-fouling operation involves effective pretreatment of feed, careful selection of the membrane and good hydrodynamics within the module, as well as an appropriate ﬂux.

References 1 Hermia, J. (1982) Transactions of the Institution of Chemical Engineers, 60, 183–187. 2 Blanpain, P. and Lalande, M. (1997) Filtration & Separation, 34, 1065–1069.

3 Field, R.W., Wu, D., Howell, J.A. and Gupta, B.B. (1995) Journal of Membrane Science, 100, 259–272. 4 Bacchin, P., Aimar, P. and Field, R.W. (2006) Journal of Membrane Science, 281, 42–69.

References 5 Belfort, G., Davis, R.H. and Zydney, A.L. (1994) Journal of Membrane Science, 96, 1–58. 6 Hong, S., Krishna, P., Hobbs, C., Kim, D. and Cho, J. (2005) Desalination, 173, 257–268. 7 McDonogh, R.M., Fane, A.G. and Fell, C.J.D. (1989) Journal of Membrane Science, 43, 69–85. 8 Chan, R., Chen, V. and Bucknall, M.P. (2004) Biotechnology and Bioengineering, 85, 190–201. 9 Lee, N., Amy, G., Croue, J.-P. and Buisson, H. (2005) Journal of Membrane Science, 261, 7–16. 10 McDonogh, R., Schaule, G. and Flemming, H.-C. (1994) Journal of Membrane Science, 87, 199–217. 11 Yamamura, H., Chae, S., Kimura, K. and Watanabe, Y. (2007) Water Research, 41, 3812–3822. 12 Fane, A.G., Wei, X. and Wang, R. (2006) Chapter 7, in Interface Science in Drinking Water Treatment (eds G. Newcombe and D. Dixon), Academic Press. 13 Chellam, S. and Xu, W. (2006) Journal of Colloid and Interface Science, 301, 248–257. 14 Van der Bruggen, B., Kim, J.H., DiGiano, F.A., Geens, J. and Vandecasteele, C. (2004) Separation and Puriﬁcation Technology, 36, 203–213. 15 Parameshwaran, K., Fane, A.G., Cho, B.D. and Kim, K.J. (2001) Water Research, 35, 4349–4358. 16 Le-Clech, P., Chen, V. and Fane, T.A.G. (2006) Journal of Membrane Science, 284, 17–53. 17 Zhang, J., Chua, H.C., Zhou, J. and Fane, A.G. (2006) Journal of Membrane Science, 284, 54–66. 18 Judd, S. (2006) The MBR book: Principles and Applications of Membrane Bioreactors in Water and Wastewater Treatment, Elsevier, Great Britain. 19 Cui, Z.F., Chang, S. and Fane, A.G. (2003) Journal of Membrane Science, 221, 1–35. 20 Le-Clech, P., Fane, A., Leslie, G. and Childress, A. (2005) Filtration & Separation, 42, 20–23.

21 Elimelech, M., Gregory, J., Jia, X. and Williams, R.A. (1995) Particle Deposition and Aggregation: Measurement, Modelling and Simulation, Butterworth-Heinemann, Woburn, Massachusetts. 22 Boerlage, S.F.E., Kennedy, M.D., Dickson, M.R., El-Hodali, D.E.Y. and Schippers, J.C. (2002) Journal of Membrane Science, 197, 1–21. 23 Schippers, J.C. and Verdouw, J. (1980) Desalination, 32, 137–148. 24 Hoek, E.M.V., Bhattacharjee, S. and Elimelech, M. (2003) Langmuir, 19, 4836–4847. 25 Brant, J.A. and Childress, A.E. (2004) Journal of Membrane Science, 241, 235–248. 26 Schwinge, J., Neal, P.R., Wiley, D.E. and Fane, A.G. (2002) Desalination, 146, 203–208. 27 Flemming, H.C. (1997) Experimental Thermal and Fluid Science, 14, 382–391. 28 Characklis, W.G. and Marshall, K.C. (1990) Bioﬁlms: A Basis for an Interdisciplinary Approach in Bioﬁlms (eds W.G. Characklis and K.C. Marshall), John Wiley & Sons Inc., New York, pp. 3–15. 29 Winters, H. (2005) International Desalination Association World Congress 2005 IDA: Singapore, pp. SP05–200. 30 Boerlage, S.F.E., Kennedy, M.D., Aniye, M.P., Abogrean, E., Tarawneh, Z.S. and Schippers, J.C. (2003) Journal of Membrane Science, 211, 271–289. 31 Kang, S.T., Subramani, A., Hoek, E.M.V., Deshusses, M.A. and Matsumoto, M.R. (2004) Journal of Membrane Science, 244, 151–165. 32 Chong, T.H., Wong, F.S. and Fane, A.G. (2008) Jounal of Membrane Science, 325, 840–850. 33 Sharma, P.K., Gibcus, M.J., van der Mei, H.C. and Busscher, H.J. (2005) Applied and Environmental Microbiology, 71, 3668–3673. 34 Hunt, S.M., Werner, E.M., Huang, B., Hamilton, M.A. and Stewart, P.S. (2004) Applied and Environmental Microbiology, 70, 7418–7425.

j137

j 6 Fouling in Membrane Processes

138

35 Wood, B.D. and Whitaker, S. (1998) Chemical Engineering Science, 53, 397–425. 36 Herzberg, M. and Elimelech, M. (2007) Journal of Membrane Science, 295, 11–20. 37 Melo, L.F. (2005) Water Science and Technology, 52, 77–84. 38 Lee, S. and Lee, C.H. (2005) Water Science and Technology, 51, 267–275. 39 Borden, J., Gilron, J. and Hasson, D. (1987) Desalination, 66, 257–269. 40 Pervov, A.G. (1991) Desalination, 83, 77–118. 41 Mullin, J.W. (1993) Crystallization, Butterworth-Heinemann, Oxford. 42 van de Lisdonk, C.A.C., Rietman, B.M., Heijman, S.G.J., Sterk, G.R. and

43

44 45

46 47 48

Schippers, J.C. (2001) Desalination, 138, 259–270. Tzotzi, C., Pahiadaki, T., Yiantsios, S.G., Karabelas, A.J. and Andritsos, N. (2007) Journal of Membrane Science, 296, 171–184. Hasson, D., Drak, A. and Semiat, R. (2003) Desalination, 157, 193–207. Da Costa, A.R., Fane, A.G., Fell, C.J.D. and Franken, A.C.M. (1991) Journal of Membrane Science, 62, 275–291. Schock, G. and Miquel, A. (1987) Desalination, 64, 339–352. Boudreau, B.P. (1996) Geochimica et Cosmochimica Acta, 60, 3139–3142. Zhang, Y.P., Chong, T.H., Fane, A.G., Law, A., Coster, H.G.L. and Winters, H. (2008) Desalination, 220, 371–379.

j139

7 Energy and Environmental Issues and Impacts of Membranes in Industry William J. Koros, Adam Kratochvil, Shu Shu, and Shabbir Husain

7.1 Introduction

Short-term economics favor rapid deployment and guaranteed reliability, so energy efﬁciency and environmental sustainability are often secondary considerations in implementation of new processes. Options with lower energy efﬁciency and higher environmental impact may be favored over membrane processes, which tend to be less familiar to design engineers and may even require some development time and risk. Moreover, while government regulations can encourage adoption of environmentally beneﬁcial approaches such as membranes, global regulations are difﬁcult to implement. Nevertheless, over the longer run, practitioners can seize a competitive advantage by moving forward ahead of regulations to deﬁne the technological landscape. In the early 1970s, Japanese automotive innovations in fuel efﬁciency enabled a very strong position to be captured in automotive production three decades later. A similar early stage opportunity exists now with regard to large-scale separations processes for production of commodity chemicals, fuels, and water. In many ways, the water-puriﬁcation sector is more advanced than the other two sectors, and understanding why the broader nonaqueous separation sector has lagged behind is useful. Such understanding provides a framework to efﬁciently extend the advantages of the membrane platform across the spectrum of separations pertinent to commodity production. While some of this information has already been reported, this chapter provides updated information and signiﬁcant expansion on the future of membrane separations [1]. Linkage exists between separation energy efﬁciency and long-term environmental sustainability, and some facts help to clarify this connection. By United Nations estimates, the world currently has 6.7 billion global inhabitants; and, only 1.2 billion people live in more-developed countries1) such as North America, western Europe, and so on, while 5.5 billion reside in less-developed countries [2]. Estimates suggest 1) A highly industrialized country characterized by signiﬁcant technological development, high per capita income, and low population growth rates.

Examples of such countries include the United States, Canada, Japan, and many countries in Europe.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

140

that 9.2 billion inhabitants will occupy our planet by 2050, and the ratio 9.2/1.2 ¼ 7.7 provides an upper bound estimate of the increased energy use relative to 2007 to provide an equivalent standard of living to all global inhabitants by 2050 [2]. Asia is currently classiﬁed as less developed, but this region has explosively growing economies with visions of joining the ranks of more-developed countries. Moreover, the UN estimates that this region will comprise a population of 5.3 billion by 2050 – over half of the worlds inhabitants! A more conservative estimate of the likely world energy consumption in 2050 might be the factor of (5.3 þ 1.2)/1.2 ¼ 5.4, or roughly 5, and this factor will be used for discussion purposes in this chapter to estimate industrial-sector use. Raising the standard of living for such a massive group requires huge increases in commodities, including clean water, nonpolluting fuels, and chemicals. From the standpoint of separation processes, providing commodities to the 5.3 billion additional inhabitants in more-developed countries by 2050 is truly a grand challenge. A useful benchmark to guide thinking with regard to separation issues is the well-studied US case where the industrial sector is responsible for 33% of total energy consumption. Over 40% of the energy consumption in the massive chemical and reﬁning and petrochemical industry is consumed by separation processes [3]. Using the scaling 0.4 0.33 ¼ 0.132, it is reasonable to estimate that 13.2% of total energy consumption can be associated with separation operations. The implications of a business as usual scaling to accommodate a projected 5 increase in global commodities would correspond to a 66% increase in current energy consumption that is associated with all aspects of the global economy in 2007. Since the bulk of such energy will come from hydrocarbon sources for many years, this energy burden would bring with it a similar increase in CO2 emissions and present a major hurdle to worldwide economic sustainability. Fortunately, membranes offer a viable option to address the separations part of this grand challenge. To be effective, however, membrane technology must be introduced prior to installation of energy inefﬁcient thermally intensive processes. Indeed, if energy-inefﬁcient process are installed, their long (30–50 year) useful lives will require regulatory intervention to force their replacement. To support the latter claim, consider production of fresh water by desalting brackish and sea water brines. Currently, there are desalting facilities worldwide with the capacity of 9 billion gallons of water per day, and roughly half is membrane-based and half is thermally driven (e.g., multi-effect and ﬂash evaporators). Since the advent of modern reverse-osmosis (RO) desalting technology, almost all new desalting capacity is based on membranes. Nevertheless, despite more than a 10 higher energy efﬁciency of membranes, which will be shown in a later section, the old thermal plants remain in use. The same situation can be expected in the chemical and petrochemical industry if scale up to handle the 5 capacity expansion is done by conventional thermally intensive approaches. This reality places added urgency on the need for expedited development of membrane-based processes that expand beyond traditional aqueous puriﬁcation of brines and micro- or ultraﬁltration of aqueous feeds. A brief review of experience with aqueous feeds will be provided to identify lessons learned that can help expedite expansion of the membrane platform to nonaqueous feeds. The subsequent discussion focuses on large-scale examples where signiﬁcant

7.2 Hydrodynamic Sieving (MF and UF) Separations

reductions in energy consumption (and hence CO2 emissions) appear feasible by replacing energy-intensive approaches with membrane processes. Although many net driving forces can be imposed on each penetrant between upstream and downstream membrane faces, transmembrane pressure differences are the most common and are the focus of this discussion. Most practical membrane processes are continuous steady-state operations with a feed, permeate, and nonpermeate stream. Since membrane processes involve separation of a permeated component A from a second, rejected component B, a measure of separation efﬁciency is useful. Due to the diversity of applications, many different measures of separation efﬁciency are used in the various membrane subareas. Probably the easiest to use measure is the so-called separation factor, given in Equation 7.1, which shows the relative enrichment of component A vs. B due to the membrane process [4]: Composition of A downstream Composition of B downstream ð7:1Þ SF ¼ Composition of A upstream Composition of B upstream Since the SF is a ratio of ratios, any measure of composition (mole fraction, mass fraction, concentration, etc.) can be used in Equation 7.1 as long as one consistently uses the same measure for both upstream and downstream phases in contact with the membrane. Locally within a module, the ratio of compositions leaving the downstream face of a membrane equals the ratio of the transmembrane ﬂuxes of A vs. B. Local ﬂuxes of each component are determined by relative transmembrane driving forces and resistances acting on each component. The ratio of the feed compositions in the denominator provides a measure of the ratio of the respective driving forces for the case of a negligible downstream pressure. This form normalizes the SF to provide a measure of efﬁciency that is ideally independent of the feed composition. For a given driving force, minimization of the membrane resistance requires the smallest possible effective membrane thickness, ‘. The ability to minimize ‘ without introducing defects relies upon micromorphology control, and this topic impacts virtually all membrane applications.

7.2 Hydrodynamic Sieving (MF and UF) Separations

Microﬁltration (MF) and ultraﬁltration (UF) involve contacting the upstream face of a porous membrane with a feed stream containing particles or macromolecules (B) suspended in a low molecular weight ﬂuid (A). The pores are simply larger in MF membranes than for UF membranes. In either case, a transmembrane pressure difference motivates the suspending ﬂuid (usually water) to pass through physically observable permanent pores in the membrane. The ﬂuid ﬂow drags suspended particles and macrosolutes to the surface of the membrane where they are rejected due to their excessive size relative to the membrane pores. This simple process

j141

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

142

concentrates particles or macromolecules in the upstream nonpermeate stream and produces essentially pure low molecular weight permeate downstream (SF ! 1) if the pore-size distribution prevents any B from passage across the membrane. Removal of a bulk liquid often represents the major energy cost for processing suspended particles and macromolecular solutes including paints, foods, and myriad waste-recycle streams. For dilute and semidilute feeds (<15 vol%), both MF and UF enable large energy savings compared to evaporation approaches [5, 6]. Despite pumping expenses to drive permeation and minimize accumulation of a rejected component at the membrane surface, energy costs typically range between 0.15–5.0 (kw h)/m3 of water removed [5, 6]. Generation of electricity using high-pressure steam gives typical efﬁciencies of 33% or less [7], thereby increasing the thermal equivalent energy cost for the membrane option. Using a median value of 2.5 (kw h)/m3 for such MF and UF membrane processes and accounting for 33% efﬁciency of steam-generated electricity, a value of (2.5/0.33) ¼ 7.6 (kw h)/m3 results. Despite this penalty, the membrane option offers roughly a 10-fold savings over competitive thermal removal [73 (kw h)/m3] by ﬂash evaporation [8]. Even if thermal energy input is needed in a ﬁnal ﬁnishing step, using membranes in primary concentration steps can provide large overall processing cost savings.

7.3 Fractionation of Low Molecular Weight Mixtures (NF, D, RO, GS)

As the size of both the permeated and rejected components become less than 20 Å, as shown in Figure 7.1, hydrodynamic sieving forces are no longer adequate to perform the subtle size and shape discrimination required. Indeed, the progressions from nanoﬁltration (NF) ! reverse osmosis (RO) ! gas separation (GS) processes represent increasingly more challenging discrimination between entities that often differ by 3–5 Å for nanoﬁltration or reverse osmosis, down to only fractions of an angstrom for gases. In all of these cases, intermolecular forces become dominant determinants of the resistance acting on each penetrant. For such micromolecularly selective processes, an additional partitioning phenomenon must also be considered in the ﬂux expression to enable describing the

Figure 7.1 Size spectrum of permeate and the controlling mechanism of transport.

7.3 Fractionation of Low Molecular Weight Mixtures (NF, D, RO, GS)

process conveniently in terms of external phase conditions. In this case, the partitioning phenomenon can be accommodated as a factor contributing to transport using a partition coefﬁcient typically deﬁned as: Ki ¼

½Composition of component i in membrane ½Composition of component i in external phase

ð7:2Þ

Since Ki is expressed as a ratio, any consistent measure of composition in the membrane and external phases may be used in Equation 7.2. When Ki > 1, the membrane acts as a concentrator that attracts component i from the external phase and makes it available at the membrane surface for transmembrane movement. Intermolecular forces of solvation and mixing that are responsible for the partitioning process may be entropic as well as enthalpic in origin. The balance of these forces acting between the membrane and external phase can cause either a higher or lower concentration of a given solute inside the membrane relative to the external phase. If the tendency to enter the membrane is negligible, the partition coefﬁcient approaches zero, that is, Ki ! 0. The synergistic action of the size-discriminating and partitioning phenomena permits adjustment of the relative compositions of different small molecules or ions in streams contacting the upstream and downstream faces of a membrane. For a given penetrant pair, the ratio of the effective resistance acting on B vs. that acting on A in the membrane speciﬁes the membrane-speciﬁc ability to separate this A–B pair. Since the thickness factor, ‘, cancels, the key ratio of resistances acting on component B vs. A is comprised of a product of partitioning and mobility ratio factors. For most membranes, the mobility ratio can be approximated as DA/DB, the ratio of the average diffusion coefﬁcients for component A vs. B within the membrane phase. In this common case, therefore the effective ideal membrane selectivity, aAB, is given by: aAB ¼

½DA ½KA ½DB ½KB

ð7:3Þ

Equation 7.3 notes that one can tune both mobility selectivity, DA/DB, and partitioning selectivity, KA/KB to develop advanced materials for every small molecule separation [9]. This strategy can be applied to virtually any type of membrane material ranging from gels to crystalline zeolites, metals, glasses, or polymers. Moreover, hybrid materials comprised of combinations of more than one such material (e.g., a zeolite dispersed in a polymer) allow limitations associated with any speciﬁc pure component material type to be overcome. For instance, intrinsic rigidity responsible for outstanding mobility selectivity in zeolites also causes brittleness and difﬁculties in their high-speed processing. Polymers are processable, but lack the rigidity to perform ﬁne mobility selectivity. Mixtures of zeolites and polymers or molecular-sieve carbons and polymers are now being investigated to create highly selective hybrid materials amenable to economical high-speed fabrication [10, 11]. As will be discussed in more detail, such materials are likely to be increasingly important for dealing with a broad range of future applications.

j143

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

144

7.4 Reverse Osmosis – The Prototype Large-Scale Success

As noted earlier, reverse-osmosis (RO) puriﬁcation of water was the ﬁrst large-scale commercially viable membrane fractionation of low molecular weight liquid mixtures. Like all of the cases involving low molecular weight fractionation, RO puriﬁcation of potable water from brine relies upon partitioning selectivity and mobility selectivity contributions from Equation 7.3. Optimization of the membrane material and structures for this application took place over a period of more than two decades, and membranes are now rapidly displacing thermal desalting [8, 12]. By understanding how and why RO has displaced distillation in this large-scale application, one can see how to help expand the energy-efﬁcient membrane paradigm more broadly. Both gas separation and RO require high feed pressures to achieve useful ﬂuxes. Nevertheless, the utility of having a high-pressure nonpermeate stream leaving the module differs greatly for GS vs. RO cases. For gases, the energy used to compress feed streams is valuable in subsequent processing and product storage. On the other hand, for liquids, after the RO separation is completed excess pressure in the nonpermeate is not needed. Reclaiming this energy is now standard procedure in state-of-the-art RO systems. Pervaporation is a variant of reverse osmosis that uses a low-pressure liquid feed with a vapor permeate under vacuum. Effectively, pervaporation involves permeation and evaporation of a portion of the feed, thereby requiring signiﬁcant thermal energy input [5]. While overcoming the need for high-pressure feed and nonpermeate energy recovery, pervaporation still requires the input of considerable thermal energy. Because reverse osmosis eliminates this thermal inefﬁciency associated with pervaporation, it became the favored process for water desalination. The effective driving force for reverse-osmosis permeation of water is proportional to the difference in applied transmembrane pressure, DP, and the transmembrane osmotic pressure, DP, viz, (DP–DP) [13]. For 50% recovery of feed entering with 34 000 ppm of total dissolved salts in seawater, the stream leaving the module has a very large osmotic pressure. This osmotic pressure must be overcome to produce the last increment of potable water product leaving the module. As noted earlier, providing a large transmembrane DP without paying an excessive energy cost is commonplace in state-of-the-art reverse-osmosis operations with compact energy recovery turbines [12]. In principle, this practice of recovery of unused energy in compressed nonpermeate streams should be transferable to organic systems as well; however, materials of construction and seals require development for compatibility with organic vs. aqueous feeds. A state-of-the-art RO seawater system processes 50 million gallons per day with 50% feedwater recovery as potable water product using a 940-psi (65 bar) feed pressure [12]. These high pressures and ﬂows are now routinely accommodated economically with compact vessels and high productivity membranes. An optimized thermal distillation plant with the same feedwater requires 1014 Btu/gal [78.5 (kw h)/m3] of water produced [8], while the state-of-the-art seawater RO system has an energy cost of only 2.2 (kw h)/m3 [8, 12]. Using the current paradigm

7.5 Energy-Efficiency Increases – A Look to the Future

of steam cycle generation of electricity with an efﬁciency of only 33%, the effective thermal equivalent energy cost for the membrane process is (2.2/0.33) (kw h)/m3 ¼ 6.7 (kw h)/m3. Again, even with such a penalty factor, the membrane option is over 10-fold more efﬁcient than the thermal approach. The well-known thermodynamic inefﬁciency in generation of electricity using highpressure steam can be linked to the unfortunate widespread acceptance of the inefﬁciency of doing thermally driven separations. Speciﬁcally, generation of electricity using high-pressure steam produces excess low-pressure steam: this fact is often used to justify continuation of inefﬁcient separation processes driven by this excess low-value steam [14]. In fact, discussions of thermal separation efﬁciencies are sometimes based on the efﬁciency of an ideal heat engine operating between the reboiler and condenser temperatures. Such an approach overlooks the intrinsic limitations of all thermally driven processes and perpetuates the unnecessary linkage between thermal energy conversion processes and separation processes. More discussion of issues related to power generation and membrane roles in reducing environmental impact will be offered later. It is well known that electrochemical oxidation of a fuel to extract power can theoretically be performed in a fuel cell much more efﬁciently than is possible via a heat cycle. For example, a H2/O2 fuel cell reaction at 25 C has an ideal efﬁciency of 100%, as compared to 30–33% in standard steam cycles. Current fuel cells still require improvement, and rarely exceed 50–60% efﬁciency; however, this already surpasses the 33% efﬁciency for standard steam systems [15]. Realistically, however, scaling such devices to hundreds of megawatt size presents challenges. Even without the ideal efﬁciency of a fuel cell, combined-cycle integrated gasiﬁcation processes are providing signiﬁcant improvements with efﬁciencies nearly as high as 50% [16, 17]. Using 50% as the efﬁciency limit, such a unit coupled to a reverse-osmosis unit would show an improvement of 73/(2.2/0.50) > 16-fold better than the thermal separation alternative! Whether one considers the already achievable 10-fold reduction with a conventional coal-ﬁred steam turbine or the 16-fold reduction achievable by eventually coupling this membrane process with a highefﬁciency integrated gassiﬁer, the numbers are impressive. These numbers also give a vision of a much more energy efﬁcient future if the membrane platform is extended to nonaqueous applications.

7.5 Energy-Efficiency Increases – A Look to the Future

The following cases consider advantages in energy savings that extend beyond aqueous ﬁltration and reverse-osmosis applications noted above. Although not yet offering a full factor of 10 savings, this is the same path toward dominance that aqueous separations followed. Almost a decade ago, the concept of a disruptive technology was introduced to describe new approaches that were radically different from the incumbent leading technology in a ﬁeld. Such technology was also noted in some cases to even perform less well than the incumbent leader but with

j145

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

146

optimization ultimately improved to become a major player, if not the dominant player in a ﬁeld. In this context, membrane processes are a potentially disruptive technology. 7.5.1 Success Stories Built on Existing Membrane Materials and Formation Technology

Valuable savings are possible even using available gas and vapor-separation membrane units, while aggressively pursuing development of nonaqueous RO and its larger energy payoffs over the next decade. Vapor-separation processes are operationally similar to gas-separation units but often use a moderate vacuum downstream, depending upon the vapor pressure of the components at the feed temperature. A number of applications have been suggested for the removal of organic vapors from gas streams. These include monomer recovery from storage-tank vent streams in the production of polyoleﬁns (e.g., polyvinyl chloride, polyethylene, polypropylene) which will be discussed in a later section [18], removal of natural-gas liquids from fuel gas for gas engines and turbines [19] and removal of solvents from air [20–22]. These applications utilize the high condensability of vapors to achieve high separation efﬁciencies between condensable and noncondensable components. Unlike separations involving permanent gases, where diffusion selectivity is the dominant factor, these membranes rely upon a so-called reverse selective process based on very high sorption selectivities to achieve separation. As an example, butadieneacrylonitrile rubber was cited as having a selectivity of around 100 000 for benzene over air [22]. Required selectivities for viablility of the separation with vapors typically lies at selectivities around 100 to 200 to minimize the gas component in the vacuum section [22]. Higher selectivities signiﬁcantly above this range provide only marginal improvement; since the economics of the separation are driven by the value of the condensable component recovered and reduction in VOC emissions. Using a rubbery membrane in combination with a ﬂash unit and condenser, recovery of up to 500–1000 lb/h of monomer and processing solvent from polymer storage bin purge waste gas streams has been reported with savings of $1 million/year/purge bin [18]. Beyond organic vapor capture applications, more standard membranes involving natural gas represent a large and attractive market for gas-separation membranes. The SACROC installation was one of the ﬁrst major applications of gas-separation membranes in large-scale separations. This application deals with removal of CO2 from natural gas associated with crude oil. The Kelly–Snyder ﬁeld was discovered in 1948 with an estimated size of 2.1 billion bbl of oil. The initial reservoir was produced using water ﬂooding that was later replaced by carbon dioxide injection by the ﬁeld operator, Chevron. A Benﬁeld (hot promoted potassium carbonate) process and amine scrubbing were employed to remove the CO2 from the associated gas in the initial stages of operations, prior to the development of membranes. An eventual increase in CO2 content of the associated gas stream from 0.5 to 40 mole% necessitated an expansion of the CO2 handling capacity. Membrane units were considered to be ideal due to their modularity that allowed easy scaling, thereby

7.5 Energy-Efficiency Increases – A Look to the Future

foregoing major capital costs associated with expanding the amine and carbonate units before the added capacity was needed. This staged increase in capacity is another particularly attractive aspect of membrane-based processes. Cellulose acetate hollow-ﬁber membranes were provided by Cynara who also operated the membrane plant, and initial testing of the membranes was carried out at the Chevron facility to increase conﬁdence in the new technology. A membrane lifespan of 5 years was reported in the presence of adequate pretreatment [23]. Since the initial deployment of membranes, the unit has been expanded from the original 70 MMscfd to 600 MMscfd in 2006 with a gas feed of 87 mole% CO2 [24, 25]. In many ways, this case delivers on the potential of membranes to be expandable to large applications beyond simple aqueous feeds. Membrane technology for natural-gas separations is gaining broad acceptance and a number of major membrane-separation plants have come into operation in recent years. These include the Cakerawala production platform (CKP) that processes 700 MMScfd with a 37% CO2 feed and a plant in Qadirpur, Pakistan that processes 500 MMscfd of 6.5 mole% CO2 feed to 2% CO2 pipeline speciﬁcation [26]. Current plans are being made to double production at the CKP facility [27]. In these applications, membranes offer the unique ability to conﬁgure compact systems to perform the desired separation. In some cases, membrane modules placed in series or in parallel enable debottlenecking, while in other cases such units can improve the overall efﬁciency of the separation. Depending on the needs of the separation, the following ﬁgures illustrate possible conﬁgurations for membrane separations. Figure 7.2 shows a simple two-stage membrane-separation process where the nonpermeate of the second stage is recycled to the feed of the ﬁrst stage. A more complex conﬁguration is presented in Figure 7.3, where both the permeate and nonpermeate streams go through a two-stage membrane process in order to achieve a higher purity of both products. An example of the compact nature and modularity of membrane units is shown in Figure 7.4 where the two membrane units in the foreground replaced the amineabsorption system in the background for removal of CO2 from natural gas. This application was mentioned earlier with regard to SACROC and subsequent larger offshore applications where space is at a premium.

Figure 7.2 Simple two-stage membrane module configuration (Adapted from Ref. [28]).

j147

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

148

Figure 7.3 Dual two-stage membrane module configuration (Adapted from Ref. [28]).

Hydrogen puriﬁcation represents another ideal ﬁt for membrane-based separations in many cases. Because of its small molecular size relative to other gases, combined high ﬂuxes and selectivities are often possible, since the diffusional selectivity in Equation 7.3 greatly favors hydrogen, even when the sorption selectivity does not. Initial deployment was carried out in ammonia plants to recover and recycle hydrogen from the product stream. This easy separation has led to a saturation of the ammonia market, with almost all units employing membranes [29]. A second more challenging, and even larger, market exists in the recovery of hydrogen in reﬁnery processes. Increasing use of heavy and sour crude oils require ever larger quantities of hydrogen for oil upgrading to adjust the carbon:hydrogen ratio for lower carbon fuel. The heavier oils are cracked down ideally to pentanes and higher but also result

Figure 7.4 Air Liquide/Medal membrane unit replacement of amine scrubbing towers (used with permission).

7.5 Energy-Efficiency Increases – A Look to the Future

in the formation of light hydrocarbons such as methane, ethane, and propane [30]. These gases act as inerts and reduce the vapor pressure of hydrogen in the reactor and must be periodically purged. Typically, 4 moles of hydrogen are lost for every mole of light hydrocarbon removed [4]. Initial deployment of membranes in hydrogen separations was driven by their exceptional payback and modular design, which allowed for their inclusion in existing reﬁnery process lines with little modiﬁcation. Replacing high-pressure purging and gas absorbers in the hydrocracking process discussed above by using membranes can decrease hydrogen losses by up to 16-fold. Depending on the process requirements of product purity, hydrogen recovery, and product pressure, the economics can justify the use of membrane technology versus traditional adsorption, cryogenic distillation and pressure swing adsorption. Examples of such comparisons are well covered by Zolandz and Fleming [4] and by Baker [30]. Temperature control of the feed stream is critical to membrane operation in order to prevent condensation of the hydrocarbons as hydrogen is removed. As the dew point is reached, the condensing hydrocarbons can lead to plasticization and membrane failure. While the feed temperature is typically kept at 15–20 C above the dew point of the retentate stream, process upsets or feed changes can still lead to membrane failure. Although as many as 100 membrane plants have been installed in reﬁneries, the global market remains far from saturation. Membranes with increased resistance to plasticization and higher-temperature operation or the use of improved pretreatment would result in greater conﬁdence on the technology and widespread adoption [30]. 7.5.2 Future Opportunities Relying Upon Developmental Membrane Materials and Formation Technology

Besides the above success stories, reconﬁguring existing thermally driven processes to produce vapor feeds to membrane units for targeted fractionations of valuable components could be an attractive evolutionary strategy. However, as economical nonaqueous RO capability develops, these processes should phase out the older thermal units in the same rapid evolutionary manner that is currently occurring in the aqueous RO arena. As noted earlier, even rapid evolution takes time (10–15 years), as it did for aqueous systems. Such a process should begin now to avoid further proliferation of additional energy-inefﬁcient separation units to meet expanding capacity needs. The following forward looking cases consider opportunities where membranes could have a large impact; however, new membranes will be needed as opposed to the previous cases where existing membranes are adequate. 7.5.2.1 High-Performance Olefin–Paraffin Separation Membranes The oleﬁns ethylene and propylene are highly important synthetic chemicals in the petrochemical industry. Large quantities of such chemicals are used as feedstock in the production of polyethylene, polypropylene, and so on [31]. The prime source of lower oleﬁns is the oleﬁn–parafﬁn mixtures from steam cracking or ﬂuid catalytic cracking in the reﬁning process [32]. Such mixtures are intrinsically difﬁcult to

j149

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

150

Table 7.1 Physical properties of ethane, ethylene, propane, and propylene.

Boiling point ( C) Lennard-Jones parameter, sLJ (Å)

Ethylene

Ethane

Propylene

Propane

Ref.

103.9 3.7

88.6 4.1

48 4.0

42.2 4.3

[33] [34]

separate due to the similar physical properties of the saturated/unsaturated hydrocarbons, as shown in Table 7.1 [33, 34]. Another important factor that distinguishes this separation is that it is not environmentally or economically feasible to simply return a rejected stream to the environment, as in a typical aqueous RO process where the brine can be returned to the ocean. The federal regulations mandate that CO2 emissions from reﬁneries and chemical plants be reduced to low levels; therefore, facilities can no longer afford to dispose of waste hydrocarbon streams in their ﬂare systems. Pure streams from polyoleﬁn reactors and vents from polymer-storage facilities, which were once ﬂared, must be redirected to recovery systems. To reduce the economic penalty of environmental compliance, these parafﬁn and oleﬁn mixtures must be recovered and recycled. In other words, two products must be made, a useful fuel and a useful chemical product, hence more process engineering is required in order to achieve such an objective. A US DOE report estimated that 1.2 1014 BTU/year are used for oleﬁn/parafﬁn separations [35]. The conventional technology to separate oleﬁn/parafﬁn mixtures is cryogenic distillation, as illustrated in Figure 7.5. The separation is performed at elevated pressures in traditional trayed fractionators. C2 and C3 distillation columns are often up to 300 feet tall and typically contain over 200 trays. Although the separation to achieve chemical grade purity can be accomplished in a single tower [36]. Purifying ethylene/propylene to polymer grade requires a signiﬁcant increase in the number of fractionating trays or the reﬂux ratio or both [36]. The large capital expense and energy cost have created the incentive to seek alternative technology for this oleﬁn/parafﬁn separation. Membranes offer excellent potential as an alternative for traditional distillation technology. A signiﬁcant amount of research has shown the potential of membranes in the oleﬁn/parafﬁn separation arena, which will be brieﬂy discussed later in this section. Capacity expansions of existing thermally driven separation units are ideal ways to introduce membranes more broadly into large-scale use while minimizing risks and building familiarity with this relatively new technology. This approach leverages existing investments without the need to build entirely new thermally driven separation units. Within an existing integrated plant, valuable compounds in a vapor feed stream currently sent to another thermally driven separation unit could be membrane-fractionated into higher value products with minimal expense and signiﬁcant energy savings. Figure 7.6 demonstrates an example of possible implementation of membrane units with C2 and C3 splitters. The introduction of membranes will lead, depending on the separation characteristics of the membrane material, to a signiﬁcant reduction of the stream fed to the splitter. A possible reduction in capacity of the splitter column might be of the highest interest because the

7.5 Energy-Efficiency Increases – A Look to the Future

Figure 7.5 Conventional cryogenic distillation process for an olefin/paraffin process (Adapted from Ref. [35]).

Figure 7.6 Conventional separation of C2 and C3 mixtures integrated with possible membrane units (Adapted from Ref. [37]).

j151

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

152

Figure 7.7 Use of a membrane unit to recover and recycle propylene to the polymerization reactor (Adapted from Ref. [38]).

oleﬁn/parafﬁn separation train is more than half of the total capital cost of an oleﬁn plant. Another example suggested by Baker is illustrated in Figure 7.7 [30]. The integration of a membrane unit into a polypropylene plant could potentially recover previously wasted propylene monomer. This membrane would remove propane, which enters as an impurity in the feed, to allow recycling of the monomer without the potentially hazardous buildup of propane. Consider, for instance, the 515 Btu/lb (0.151 kw h/lb) reboiler energy is required for the propylene/propane separation using cryogenic distillation [3]. With a typical 50/50 feed and recovery of a 99.5% propylene product, this corresponds to roughly 0.302 kw h/lb propylene product. A recent patent on a vapor permeation membrane cites an energy cost of roughly 0.050 kw h/lb propylene product for this separation with a membrane having intrinsic properties similar to those currently reported in the literature [28]. As in the water RO case, accounting for the current paradigm of steam-cycle-generated electricity with a typical efﬁciency of only 33% gives the effective thermal equivalent energy cost of (0.05/0.33)Btu/gal ¼ 0.151 Btu/gal – still greatly superior to the thermal option. Moreover, as in the RO example, integrating such a process with a 50% efﬁcient fuel cell or combined-cycle gasiﬁcation process shows an improvement of 0.302/(0.05/0.5) > threefold better than the thermal alternative! First-generation membranes have been reported with properties that suggest this separation can be achieved, so this type of application is likely to develop over the next few years [39]. The ﬁrst-generation membranes investigated include polymeric membranes and polymer/silver salt composite membranes. Polymers such as cellulose acetate, polysulfone, PDMS, and polyethylene show very poor separation-performance

7.5 Energy-Efficiency Increases – A Look to the Future

Figure 7.8 C3H6/C3H8 experimental upper bound based on pure gas permeation data over the range 1–4 atm feed pressure. & ¼ 100 C, & ¼ 50 C, . ¼ 35 C, ~ ¼ 30 C, ^ ¼ 26 C (Adapted from Ref. [38]).

stability for oleﬁn/parafﬁn separation, with selectivities often below 3 [40–42]. Polyimides seem to be the most promising membrane material for this particular separation, yet the performance of polyimides was discovered to be limited by an upper-bound trade-off curve displayed in Figure 7.8 [38]. The permeability and selectivity are in reverse proportion to one another while the commercially attractive region corresponds to the upper right corner of the plot. Another category of studies focuses on facilitated transport using ion-exchange membranes that contain silver ions as a complexing agent [43–45]. Membranes with facilitated transport properties show very good selectivity and relatively high permeability coefﬁcients for the oleﬁns, but the separation process typically has to be carried out under saturated water vapor to ensure the transport of silver/oleﬁn complexes through the membrane, which adds considerable complications. Moreover, the stability of the silver carrier poses another challenge for industrial application of this technology. In addition to the polymer and facilitated transport membranes, novel materials are being proposed and investigated to achieve membranes with economically attractive properties. Carbon molecular sieve (CMS) membranes prepared by pyrolysis of polyimides displayed much better performance for oleﬁn/parafﬁn separation than the precursor membranes [39, 46, 47]. Results obtained with CMS membranes indicated properties well beyond the upper-bond trade-off curve, as shown in Figure 7.8. Nonetheless, this class of materials is very expensive to fabricate at the present time. An easy, reliable, and more economical way to form asymmetric CMS hollow ﬁbers needs to be addressed from a practical viewpoint.

j153

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

154

The preceding discussions illustrate that membranes have shown great potential as an alternative for oleﬁn/parafﬁn separation, yet the performance of current membranes is insufﬁcient for commercial deployment of this technology. Advanced material development is highly desired to improve the membrane properties and reduce cost. Another possible approach involves hybrid membranes with zeolites or CMS incorporated in a continuous polymer phase. More discussion in this regard will be covered later in this chapter. 7.5.2.2 Coal Gasification with CO2 Capture for Sequestration Membranes can contribute signiﬁcantly to new concepts in more energy-efﬁcient and low CO2 emission power generation, and the following section explores some of these cases as alternatives to conventional amine-absorption-based thermally driven processes. A state-of-the-art gasiﬁer with integrated gasiﬁcation combined cycle (IGCC) power plant, shown in Figure 7.9, enables the efﬁcient use of coal for power generation. CO2 is typically captured following gasiﬁcation and a water gas shift reaction and prior to syngas combustion in the gas turbine in an IGCC power plant. The water gas shift reactor converts nearly all CO produced during gasiﬁcation to CO2. Therefore, the CO2 concentration in the syngas leaving the shift reactor is typically in the range of 15–60% (dry basis) with total gas pressures ranging from 300 to 400 psia [16, 17, 48, 49]. This precombustion CO2 capture of the pressurized syngas is typically less costly than postcombustion CO2 capture, which requires treatment of large volumes of gas near atmospheric pressure. State-of-the-art precombustion CO2 capture technology in an IGCC plant employs amine-absorption treatment of the syngas; however, even when optimized, this

Figure 7.9 IGCC schematic with air or O2 blown gasifier (Adapted from Ref. [48]).

7.5 Energy-Efficiency Increases – A Look to the Future

treatment adds a great deal of cost to the process. In fact, depending on whether an air-blown or oxygen-blown gasiﬁer is used, gasiﬁcation that utilizes amine absorption for CO2 capture has 71–89% higher capital costs than without CO2 capture [48]. Unfortunately, a single-stage polymer membrane unit that performs a similar separation following oxygen-blown gasiﬁcation has capital costs 105% higher than without CO2 capture [16], however, as will be discussed next, membrane separation units have lower operational and environmental costs that can offset this higher current capital cost. Incorporation of carbon-capture technology also reduces the overall efﬁciency of the power plant since the process requires energy to operate. The overall efﬁciency loss for a membrane capture process in an IGCC is comparable, 9.9–13.5% drop in lower heating value (LHV), to that of the amine-capture process described above, 8.1–15% drop (LHV) [16, 49–51]. Ultimately, when considering different carbon-capture technologies, it is often best to compare the cost per ton of CO2 avoided, which takes into account the costs associated with the equipment and operation as well as the amount of CO2 removed from emissions. The amineabsorption capture technology for the oxygen-blown gasiﬁcation process has an estimated cost of CO2 avoided of $48.3/ton; whereas, membrane capture technology for the same gasiﬁcation process has an estimated cost of CO2 avoided of only $41–47/ton [16, 17, 49]. While the initial capital investment of the precombustion membrane-separation process may be slightly more expensive, there are many aspects of membranes along with the lower cost of CO2 avoided that make them more advantageous than the traditional amine-absorption process. The state-of-the-art amine-treatment approach described in the above air blown IGCC work [16] is similar to the approach that has been implemented by Dakota Gasiﬁcation at the Great Plains Synfuels Plant [52]. This process utilizes air as the oxidant with a sub-bituminous Powder Ridge Basin (PRB) coal, and the amine system operates with the following complex ﬂow sheet shown in Figure 7.10 [16]. The many units and piping are shown to illustrate the process complexity. The feed to the amine scrubbing system is pretreated with highefﬁciency microﬁltration membranes, referred to as particle-capture devices, to eliminate particulates. Also, the temperature of the stream has been reduced from 550 F (287 C) to only 100 F (38 C) by efﬁcient steam-cycle condensate and cooling water [16]. Despite the effectiveness of the amine-based system in Figure 7.10, the amine approach adds considerable complexity to the ﬁnal power system. In addition, while the amine solvent used in this process (MDEA) is considered the most stable and efﬁcient solvent for a high CO2 concentration stream, it still degrades in the presence of oxygen or when subjected to high temperatures, as found in the regeneration boiler, and can lead to corrosion of the equipment [53]. The pretreatments, described above, that deliver a particulate-free stream at 38 C to the amine system provide a ready-made feed for processing via membrane modules. This feed can be used with simple and efﬁcient membranes, new structured sorbents, membrane þ structured sorbent hybrid systems or more advanced super H2 selective membranes. These membrane systems can simplify and condense the ﬂow sheet in Figure 7.10, thereby enabling a more compact plant with less piping and associated maintenance concerns.

j155

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

Figure 7.10 Amine process for CO2 capture [16].

156

7.5 Energy-Efficiency Increases – A Look to the Future

Figure 7.11 Dual-membrane separation utilizing a highly selective metal or inorganic membrane for H2 purification and a conventional polymer membrane for the CO2/N2 separation.

The following example illustrates the potential of membrane-separation processes for precombustion carbon capture in an IGCC. This approach avoids using an expensive air-separation unit (asu) or a difﬁcult-to-implement high-temperature mixed-ion conducting membrane process; however, it still enables capture of CO2 at purities suitable for commercial use or sequestration. Figure 7.11 shows a simpliﬁed ﬂow sheet of two membrane separation units where the ﬁrst stage comprises of a zeolite, palladium, or zeolite-ceramic highly selective H2 membrane and the second stage comprises a conventional polymer membrane having the ability to reject N2 vs. CO2. There currently exist high-performance polymer membranes capable of performing the required separation in the second membrane stage. Existing polymer membranes have selectivities for CO2 vs. N2 as high as 60 at 35 C [9, 54–56]. Moreover, one should be able to tailor the properties of various polymer families used to form membranes (e.g., polyimides, polysulfones) to enable tuning of the CO2/N2 permselectivity by adding groups with a favorable interaction with CO2. It is fully expected, therefore, that an economical polymer with a N2/CO2 selectivity above 30–40 could be achieved for a feed at roughly 400 psia. In addition, this second-stage membrane would greatly beneﬁt from the high CO2 driving force at 400 psia after H2 is removed as a permeate in the ﬁrst stage. For this unusual application, the more condensable nature of CO2 and its smaller size, relative to N2, makes both factors in Equation 7.3 favor CO2 vs. N2. This favorable fact enables the desired high permselectivity of CO2 relative to N2. This situation will result in the vast majority of the CO2 permeating through the membrane to the lowerpressure side, while most of the N2 would be rejected at the high-pressure side. The high-pressure retentate stream will have low levels of CO2 and H2 and comprise a large ﬂow of mostly inert N2. This stream could also be expanded in a gas turbine to claim considerable useful work and be used for other purposes as well. The CO2 could then be compressed for transport and storage without the added volume, and ultimately cost, of nitrogen. In fact, the energy captured by expansion of the nitrogen retentate would help provide some of this required compression energy for the CO2. The main challenge of the ﬁrst separation involves development of a viable membrane. An economical highly H2 selective membrane with the ability to reject both N2 and CO2 is required for this stage, and such a membrane does not yet exist. Polymer–zeolite or ceramic–zeolite hybrid membranes may provide the required

j157

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

158

selectivity; however, a palladium, palladium alloy or pure zeolite membrane may be needed to achieve this very difﬁcult goal. This factor certainly presents a major hurdle to application of this dual-membrane system, since the cost of either the metal or zeolite membranes will be considerable. If this hurdle can be overcome, the H2 permeate stream could then be mixed with air and used in a standard combustion cycle where N2 will moderate the combustion to avoid excessive temperatures that would damage the turbine or require exotic materials of construction. A potential advantage is that the ﬁrst-stage process could operate efﬁciently at elevated temperature allowing cooling to be deferred until after H2 removal. This ﬁrst-stage membrane will require considerably more development time to implement than the second-stage membrane; however, it has the potential to be a revolutionary purely membrane-based technology for H2 production and CO2 removal.

7.6 Key Hurdles to Overcome for Broadly Expanding the Membrane-Separation Platform

The previous examples for large-scale gas and vapor separations noted above illustrate that much more advanced but still economical membranes are required to better expand the membrane platform. Three related hurdles exist to broadly extending existing membrane separation successes to other low molecular weight organic compounds: these hurdles are the lack of economical materials, membranes, and module fabrication methods. This is a serious situation that must be addressed with integrated programs that seek to develop high-efﬁciency module formation, high-speed processing, micromorphology control, and advanced materials for membrane implementation. Indeed, large osmotic pressures, higher temperatures, and more aggressive organic feeds in these systems will require even more robust membranes and modules than are currently available for water feed streams. For instance, at 25 C an osmotic pressure of roughly 79 atm (1161 psi) must be overcome to cause forward ﬂux of propylene from a 75/25 molar mixture to produce a 95/5 molar downstream mixture of propylene and propane. Such pressures can even now be contained within a compact membrane vessel, and some gas-separation modules already operate with higher feed pressures. Such a liquid RO system would even further increase the energy savings below those cited for the propylene/propane vapor-separation case mentioned above. Similarly, a low-cost palladium or pure zeolite membrane that allows only passage of H2 could enable the ﬁrst-stage membrane for hydrogen puriﬁcation in the gasiﬁcation example noted above. Much of the technology for gas- and vapor-separation materials, membranes, and modules that are now emerging as large-scale units were derived from work supported on reverse osmosis in the early 1960s [57]. This program by the Ofﬁce of Saline Water (OSW) targeted energy-efﬁcient processes based upon the promising but unproven (at the time) membranes for aqueous separation. These early membranes were, at that time, in a similar state to those for current organic systems, and many problems had to be overcome [13, 57]. While sharing some aspects with aqueous feeds, nonaqueous feeds present new challenges that must be attacked

7.6 Key Hurdles to Overcome for Broadly Expanding the Membrane-Separation Platform

Figure 7.12 Advanced materials spectrum, including not only the extremes of organic polymers and inorganic or carbon materials, but also hybrids of these materials to provide property and processing advantages.

holistically within the framework of the previously stated programs for membrane implementation. A program similar to the OSW initiative, but aimed at organic feeds and high-pressure gases and vapors, would be a positive step in this direction and an investment in the future. Such a sea-changing undertaking probably requires government initiation, as it did in the visionary OSW case. Despite demanding requirements for selectivity and robustness in this next generation of applications, membranes and modules must retain their attractive cost advantages. Realistically, therefore, any program to introduce truly new highperformance membranes should incorporate hybrid materials within its enabling vision. A complete picture of membrane materials includes the spectrum ranging from purely inorganics and carbons to purely organic polymers shown in Figure 7.12. Current work has really only explored the two extreme ends of this spectrum, plus a few hybrids containing 10–15 vol% inorganic or carbon-dispersed phases in a polymer continuous phase. For future demanding applications, it is likely that the optimum position in the materials spectrum in Figure 7.12 may be even past the midpoint in hybrid composition. Indeed, very high percentages of inorganic or carbon solids, compatibly bound within an appropriate polymer matrix, could be the preferred membrane material of the future for many applications. Such hybrids have the potential to provide the selectivity and strength of inorganics and carbons and the processability and ﬂexibility of polymers. While silane-treated zeolites dispersed in a polymer matrix have been reported to possess excellent performance in dense ﬁlms [58], the performance has been difﬁcult to replicate in asymmetric hollow-ﬁber membranes. Recently, considerable success has been achieved in approaching the major hurdle of zeolite/polymer interface by a novel route [59–63], which overcomes the limitation of silane-coupling agents observed in the phase-separating environment of asymmetric membrane formation. The approach, using an acid halide and a Grignard reagent, modiﬁes the surface of zeolite particles to increasesurface roughnessand has beensuccessfullyemployed tomodifythe surface of two small-pore zeolites, SSZ-13 and zeolite A. These modiﬁed zeolites were found to form strong adhesion with high glass transition polymers such as Ultem 1000 polyetherimide and Matrimid 5218 polyimide in dense ﬁlm and asymmetric hollowﬁber membranes, thereby providing superior gas-separation performance. Figure 7.13 displays a scanning electron micrograph of a dual-layer hollow ﬁber composed of an Ultem 1000 polymer matrix with 10 wt% Grignard-treated submicrometer zeolite A. The inset focuses on the skin region of the ﬁber showing the homogeneous dispersion of the modiﬁed zeolite particles and the excellent adhesion with the polymer matrix.

j159

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

160

Figure 7.13 SEM images of a dual-layer hollow fiber with zeolite insert. Enlargement reveals good polymer adhesion to the zeolite particles (Adapted from Ref. [62]).

The above laboratory-based successes show the future of this technology is promising, but they need to be supplemented with novel modiﬁcation and processing techniques that can be scaled for high production levels. These hybrid membranes show considerable potential, however, they still require extensive research before implementation. Indeed if the number of patents ﬁled recently is any indication, hybrid membranes are attracting industrial attention [64–81].

7.7 Some Concluding Thoughts

The above illustrations show that major, even revolutionary, energy savings are possible relative to competitive, thermally driven options by introducing membrane processes for separations. Nevertheless, the discussion also clariﬁes the need for a large-scale integrated systematic approach to greatly broaden the economical application of membranes to more aggressive feed streams. This information highlights the need for modeling and analysis that starts at megascale plant systems and ranges down to the molecular scale where most separations ultimately occur. Materials science is a critical component; however, technologies to engineer supermolecular membrane morphologies and economical modules are equally critical to build such an expanded platform. In addition to its central role in advanced separation devices considered here, aspects of membrane technology indirectly impact fuel cells, advanced batteries used in hybrid vehicles, and low-cost ﬂexible solar-energy cells. Applying all of these related energy-saving devices across the various sectors of society mentioned in the introduction of this chapter would motivate rational change toward energy efﬁciency. The special opportunities for synergistic combination of fuel cells and membrane-separation technologies should be vigorously pursued to break the unnecessary current linkage between inefﬁcacies in thermal energy-conversion

References

processes and separation processes. In order to move toward green energy processes, a high-proﬁle effort is underway to promote the introduction of fuel cells, advanced batteries, and solar cells; however, much less aggressive action is apparent to promote energy-efﬁcient separations. A concerted effort focused on developing the membrane platform beyond its current state to enable rapid replacement of energyinefﬁcient separation processes is greatly needed. It is extremely important for developing countries with fewer installed thermal processes to make investments in these more-efﬁcient approaches, and for more-developed countries to phase out these thermal processes. Ultimately, an economy based on thermal dinosaurs stands to be the biggest loser – natural selection works!

Acknowledgments

Support is acknowledged from the Georgia Research Alliance and the DOE, under grant DE-FG03-95ER14 538, and GRP Investigator grant KUS-l1-011-21.

References 1 Koros, W.J. (2004) Evolving beyond the thermal age of separation processes: membranes can lead the way. AIChE Journal, 50 (10), 2326. 2 United Nations (2007) Department of Economic and Social Affairs, Population Division. World Population Prospects: The 2006 Revision, Highlights, Working Paper No. SA/P/WP.202. 3 Humphrey, J.L. and Keller, G.E. (1997) Energy Considerations, in Separation Process Technology, McGraw-Hill, New York, NY, Chapter 6. 4 Ho, W.S. and Sirkar, K.K. (1992) Overview, in Membrane Handbook, Van Nostrand Reinhold, New York, NY, Chapter 1. 5 Eykamp, W. (1997) Membrne Separation Processes, in Perrys Chemical Engineers Handbook, 7th edn, McGraw Hill, New York, NY, Chapter 22. 6 Blume, I.(May 2004) Norit Ultraﬁtration as Pretreatment for RO for Wastewater Reuse: The Sulaibiya Project. Presentation at Advanced Membrane Technology II Conference, Irsee, Germany, Engineering Conferences International, organizer N.N. Li.

7 Smith, J.M., van Ness, H.C. and Abbot, M.M. (2001) Production of Power from Heat, in Introduction to Chemical Engineering Thermodynamics, 6th edn, McGraw Hill, New York, NY, Chapter 8. 8 Pankratz, T. and Tonner, J. (2003) Desalination.com: An Environmental Primer, Lone Oak Publishing, Houston, TX. 9 Koros, W.J., Fleming, G.K., Jordan, S.M., Kim, T.H. and Hoehn, H.H. (1988) Polymeric membrane materials for solution-diffusion based permeation separations. Progress in Polymer Science, 13, 339–401. 10 Ekiner, O.M. and Kulkarni, S.S.(December 2003) Process for Making Mixed Matrix Hollow Fiber Membranes for Gas Separation, US Patent 6,663,805. 11 Vu, D., Koros, W.J., Mahajan, R. and Miller, S.J. (January 2003) Gas Separation Using Mixed Matrix Membranes. US Patent 6,503,295. 12 Marks, D.H., Balaban, M., Falagan, B.A., Jacangelo, J.G., Jones, K.L., Koros, W.J., Letey, J., Pankratz, T.M., Sakaji, R.H., Turner, C.D. and Wilf, M. (2004) Review of

j161

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

162

13

14

15

16

17

18

19

20

21

22

the Desalination and Water Puriﬁcation Technology Roadmap, National Research Council, National Academies Press, Washington, DC. Merten, U. (1966) Transport Properties of Osmotic Membranes, in Desalination by Reverse Osmosis, M.I.T. Press, Cambridge, MA, Chapter 2. Steinmeyer, D. (1997) Energy Conservation in Separation Processes, in Encyclopedia of Separation Technology, John Wiley & Sons, Ltd, New York, NY, pp. 749–759. Brandon, N.P., Skinner, S. and Steele, B.C.H. (2003) Recent advances in materials for fuel cells. Annual Reviews of Material Science, 33, 183–213. Kaldis, S.P., Skodras, G. and Sakellaropoulos, G.P. (2004) Energy and capital cost analysis of CO2 capture in coal IGCC processes via gas separation membranes. Fuel Processing Technology, 85 (5), 337–346. Metz, B., Davidson, O., de Coninck, H.C., Loos, M. and Meyer, L.A. (2005) IPCC Special Report on Carbon Dioxide Capture and Storage, Cambridge University Press, Cambridge, United Kingdom, and New York, NY, USA. Gottschlich, D. and Jacobs, M.L. (1998) Monomer Recovery Process, Membrane Technology and Research Inc., USA, p. 14. Lokhandwala, K.A. and Jacobs, M.L. (2000) Membranes for fuel gas conditioning. Hydrocarbon Engineering, 5 (5), 81–82, 84. Strathman, H., Bell, C.M. and Kimmerle, K. (1986) Development of synthetic membranes for gas and vapor separation. Pure and Applied Chemistry, 58 (12), 1663–1668. Peinemann, K.V., Mohr, J.M. and Baker, R.W. (1986) The separation of organic vapors from air. AIChE Symposium Series, 82 (250), 19–26. Baker, R.W., Yoshioka, N., Mohr, J.M. and Kahn, A.J. (1987) Separation of organic vapors from air. Journal of Membrane Science, 31 (2–3), 259–271.

23 Juyal, U. (2002) CO2 membrane Technology Matures. Oil & Gas Journal, 100 (15), 46. 24 Parro, D. (1984) Membrane carbon dioxide separation proves out at Sacroc tertiary recovery project. Oil & Gas Journal, 82 (39), 85–86, 88. 25 Marquez, J. Jr and Brantana, M. (2006) High-capacity gas membrane elements reduce weight, cost. Oil & Gas Journal, 104 (28), 51–54. 26 UOP (29 March 2004) Expanded OGDCL Membrane Plant Meets UOPs Revamp Guarantees, Press Release. 27 Callison, A. and Davidson, G. (2007) Offshore processing plant uses membranes for CO2 removal. Oil & Gas Journal, 105 (20), 56–58, 60, 62, 64–65. 28 Collings, C.W., Huff, G.A. and Bartels, J.V. (January 2004) Separation Processes Using Solid Perm-Selective Membranes in Multiple Groups for Simultaneous Recovery of Speciﬁed Products from a Fluid Mixture. US Patent Appl. Publ. 20040004040 A1. 29 Nunes, S.P. and Peinemann, K.V. (2006) Membrane Technology: In the Chemical Industry, Wiley-VCH, Weinheim, Germany. 30 Baker, R.W. (2002) Future Directions of Membrane Gas Separation Technology. Industrial & Engineering Chemistry Research, 41 (6), 1393–1411. 31 Gottschlich, D.E. and Roberts, D.L. (1990) Energy Minimization of Separation Processes using Conventional/Membrane Hybrid Systems, Department of Energy, Report No. DE-AC-07-76ID01570. 32 (June 2001) Ethylene/Propylene, Nexant Chem Systems Report. 33 Perry, R.H. and Green, D. (1984) Perrys Chemical Engineers Handbook, 6th edn, McGraw-Hill, New York. 34 Hirschfelder, J.H., Curtiss, C.F. and Bird, R.B. (1964) Molecular Theory of Gases and Liquids, John Wiley & Sons, Ltd, New York. 35 Eldridge, R.B. (1993) Oleﬁn parafﬁn separation technology - a review. Industrial

References

36

37

38

39

40

41

42

43

44

45

& Engineering Chemistry Research, 32, 2208–2212. Humphrey, J.L., Seibert, A.F. and Koort, R.A. (1994) Separation Technologies Advances and Opportunities, Department of Energy, Report No. DE-AC-0790ID12920. Staudt-Bickle, C. and Koros, W.J. (2000) Oleﬁn/parafﬁn gas separations with 6FDA-based polyimide membranes. Journal of Membrane Science, 17, 205–214. Burns, R.L. (2002). Investigation of Poly (pyrrolone-imide) Materials for the Oleﬁn/Parafﬁn Separation, PhD Dissertation, The University of Texas at Austin. Yoshino, M., Nakamura, S., Kita, H., Okamoto, K.-i., Tanihara, N. and Kusuki, Y. (2003) Oleﬁn/parafﬁn separation performance of asymmetric hollow ﬁber membrane of 6FDA/BPDA-DDBT copolyimide. Journal of Membrane Science, 212, 13–27. Ito, A. and Hwang, S.T. (1989) Permeation of propane and propylene through cellulosic polymer membranes. Journal of Applied Polymer Science, 38, 483. Tanaka, K., Taguchi, A., Hao, J., Kita, H. and Okamoto, K. (1996) Permeation and separation properties of polyimide membranes to oleﬁns and parafﬁns. Journal of Membrane Science, 121, 197. Shimazu, A., Ikeda, K. and Hachisuka, H. (1998) Method of selectively separating unsaturated hydrocarbon, US Patent 5,749,943. Ho, W.S. and Dalrymple, D.C. (1994) Facilitated transport of oleﬁns in Ag þ containing polymer membranes. Journal of Membrane Science, 91, 13. Yang, J.S. and Hsiue, G.H. (1997) Selective oleﬁn permeation through Ag(I) contained silicone rubber-graft-poly(acrylic acid) membranes. Journal of Membrane Science, 126, 139. Yamaguchi, T., Baertsch, C., Koval, C.A., Noble, R.D. and Bowman, C.N. (1996) Oleﬁn separation using silver impregnated ion-exchange membranes and silver

46

47

48

49

50

51

52

53

54

55

salt/polymer blend membranes. Journal of Membrane Science, 117, 151. Suda, H. and Baraya, K. (1997) Alkene/alkane permselectivities of a carbon molecular sieve membrane. Chemical Communications, 93. Hayashi, J., Mizuta, H., Yamamoto, M., Kusakabe, K. and Morooka, S. (1996) Separation of ethane/ethylene and propane/propylene systems with a carbonizad BPDA-ppODA polyimide membrana. Industrial & Engineering Chemistry Research, 35, 4176. Bonsu, A.K., Eiland, J.D., Gardner, B.F., Powell, C.A., Rogers, L.H., Booras, G.S., Breault, R.W. and Salazar, N. (2006) Impact of CO2 Capture on Transport Gasiﬁer IGCC Power Plant. International Technical Conference on Coal Utilization and Fuel Systems, Clearwater, FL, May 21–25. Grainger, D. and H€agg, M.-B. (2007) Techno-economic evaluation of a PVAm CO2-selective membrane in an IGCC power plant with CO2 capture. Fuel, 81 (8), 14–24. Davison, J., Bressan, L. and Domenichini, R. (October 2003) Coal Power Plants with CO2 Capture: The IGCC Option, Gasiﬁcation Technologies, San Francisco, California. Kanniche, M. and Bouallou, C. (2007) CO2 capture study in advanced integrated gasiﬁcation combined cycle. Applied Thermal Engineering, 27, 2693–2702. Perry, M. and Eliason, D. CO2 Recovery and Sequestration at Dakota Gasiﬁcation Company. (2004) Gasiﬁcation Technologies Conference, San Francisco, California. Korens, N., Simbeck, D. and Wilhelm, D. (2002) Process Screening Analysis of Alternative Gas Treating and Sulfur Removal for Gasiﬁcation, Revised Final Report, SFA Paciﬁc, Inc. Bixler, H.J. and Sweeting, O.J. (1971) The Science & Technology of Polymer Films, vol. II, John Wiley & Sons, Ltd, New York. Koros, W.J., Coleman, M.R. and Walker, D.R.B. (1992) Controlled permeability

j163

j 7 Energy and Environmental Issues and Impacts of Membranes in Industry

164

56

57

58

59

60

61

62

63

64

65

polymer membranes. Annual Review of Materials Science, 22, 47–89. Powell, C.E. and Qiao, G.G. (2006) Polymeric CO2/N2 gas separation membranes for the capture of CO2 from power plant ﬂue gases. Journal of Membrane Science, 279, 1–49. Lonsdale, H.K. (1982) The Growth of Membrane Technology. Journal of Membrane Science, 10, 81–181. Mahajan, R. and Koros, W.J. (2002) Mixed matrix membrane materials with glassy polymers. Part 2. Polymer Engineering and Science, 42 (7), 1432–1441. Shu, S., Husain, S. and Koros, W.J. (2007) Formation of nanostructured zeolite particle surfaces via a halide/Grignard route. Chemistry of Materials, 19 (16), 4000–4006. Husain, S. and Koros, W.J. (2007) Mixed matrix hollow ﬁber membranes made with modiﬁed HSSZ-13 zeolite in polyetherimide polymer matrix for gas separation. Journal of Membrane Science, 288 (1–2), 195–207. Shu, S., Husain, S. and Koros, W.J. (2007) A general strategy for adhesion enhancement in polymeric composites by formation of nanostructured particle surfaces. Journal of Physical Chemistry C, 111 (2), 652–657. Husain, S. (2006). Mixed Matrix Dual Layer Hollow Fiber Membranes for Natural Gas Separation, in Chemical & Biomolecular Engineering, PhD Dissertation, Georgia Institute of Technology. Shu, S. (2007) Engineering the Performance of Mixed Matrix Membranes for Gas Separations, in Chemical & Biomolecular Engineering, PhD Dissertation, Georgia Institute of Technology. Koros, W.J. et al. (2002) Mixed matrix membranes and manufacture and puriﬁcation of methane from a gas stream, (USA). Application: US. 2002056369. Koros, W.J. et al. (2002) Mixed matrix membranes with pyrolyzed carbon sieve particles and methods of making and

66

67

68

69

70

71

72

73

74

75

76

using the same in synthesis gas puriﬁcation, (Chevron USA Inc., USA; The University of Texas System). Application: WO. 2002024310. Kulkarni, S.S., Ekiner, O.M. and Hasse, D.J. (2005) Method for forming of mixed matrix composite permselective ﬂuid sepn. membrane using washed mol. sieve particles, (USA). Application: US 20050230305. Kulkarni, S.S. and Hasse, D.J. (2005) Molecular sieve-containing polyimide mixed matrix membranes, (USA). Application: US. 20050268782. Kulkarni, S.S. et al. (2006) Making mixed matrix membranes using electrostatically stabilized suspensions, membranes for ﬂuid separation, and separating a ﬂuid from ﬂuid mixture, (USA). Application: US 20060117949. Koros, W.J. et al. (January 2003) Gas separations using mixed matrix membranes. US Patent 6,503,295. Miller, S.J. et al. (December 2002) Puriﬁcation of p-xylene using composite mixed matrix membranes. US Patent 6,500,233. Koros F W.J. et al. (July 2003) Mixed matrix membranes and methods for making the same. US Patent 6,585,802. Hasse, D.J. et al. (September 2003) Mixed matrix membranes incorporating chabazite type molecular sieves. US Patent 6,626,980. Ekiner, O.M. and Kulkarni, S.S.(December 2003) Process for making hollow ﬁber mixed matrix membranes. US Patent 6,663,805. Kulkarni, S.S. et al. (January 2003) Gas separation membrane with organosilicontreated molecular sieve. US Patent 6,508,860. Koros, W.J. et al. (June 2004) Crosslinked and crosslinkable hollow ﬁber mixed matrix membrane and method of making same. US Patent 6,755,900. Kulprathipanja, S. and Charoenphol, J. (April 2004) Mixed matrix membrane for separation of gases. US Patent 6,726,744.

References 77 Guiver, M.D. et al. (August 2003) Composite gas separation membranes. US Patent 6,605,140. 78 Marand, E. et al. (September 2006) Mixed matrix membranes. US Patent 7,109,140. 79 Miller, S.J. et al. (November 2006) Mixed matrix membranes with low silica-toalumina ratio molecular sieves and methods for making and using the membranes. US Patent 7,138,006.

80 Miller, S.J. et al. (January 2007) Mixed matrix membranes with small pore molecular sieves and methods for making and using the membranes. US Patent 7,166,146. 81 Miller, S.J. and Yuen, L.-T.(September 2007) Mixed matrix membrane with super water washed silica containing molecular sieves and methods for making and using the same. US Patent 7,268,094.

j165

j167

8 Membrane Gas-Separation: Applications Richard W. Baker

8.1 Industry Background

Gas separation with membranes is now a US$300–400 million dollar-per-year industry, and the industry is growing at a double-digit rate [1–3]. In the early 1980s, the ﬁrst successful membrane-based industrial gas-separation plants were built by Permea (now a division of Air Products). These plants recovered hydrogen from ammonia reactor purge gas or puriﬁed hydrogen recycle streams in reﬁnery hydrocrackers. However, the industry really took off a few years later when Medal, Generon, Ube, and Permea developed membranes to separate nitrogen from air. Nitrogen production now represents half of the membrane gas-separation equipment business. The average nitrogen separation unit cost is small, usually in the range of US$10 000–100 000, but several thousand are made each year. Another major application of gas-separation membranes is the separation of carbon dioxide from natural gas. In contrast to nitrogen units, carbon-dioxide separation plants are often very large, and cost from US$20 to 50 million each. More than twenty of these large plants have now been installed and this application continues to grow rapidly. A table describing the current membrane gas-separation industry is given below (Table 8.1). 8.2 Current Membrane Gas-Separation Technology

All gas-separation membranes have an anisotropic structure with a thin, dense selective layer facing the high-pressure feed gas. The selective layer is supported on a much thicker microporous support layer that provides mechanical strength. The chemical structure determines the permeability of the selective layer.1) The selective1) Permeability is the general term used to describe the rate at which a gas will pass through a material. Permeability is most commonly measured in Barrer, deﬁned as 1 1010 cm3 cm3 (STP) cm/cm2 s cmHg. Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 8 Membrane Gas-Separation: Applications

168

Table 8.1 Characteristics of the current membrane gas separations industry – 2008.

Separation Nitrogen from air Water from air

Carbon dioxide from natural gas

Reﬁning: H2/CH4 Ammonia plants: H2/N2, Ar Syngas: H2/CO C3 þ Hydrocarbons/ nitrogen

Principal producers 9 > Permea > > > > =Medal Dow, Generon > > > >Ube > ; Aquilo 9 > Cynara > > > > = Medal > > > Grace, Separex > > ; MTR 9 > Permea > > = Medal > > > ; MTR/Ube 9 > =MTR > ;Borsig

9 Everything else: Nitrogen/ > > > > > natural gas, Helium/ > > = natural gas, H2S/natural gas, CO2/H2, > > > > > Miscellaneous > > ; petrochemicals

—

Membrane/ modules

Market size (US$million/y)

Polysulfone/capillary ﬁber 150 Polyimide/capillary ﬁber Polyimide/capillary ﬁber Polyimide/capillary ﬁber Polyphenylene oxide/capillary ﬁber Cellulose triacetate/ ﬁne ﬁber Polyaramide/ﬁne ﬁber Cellulose acetate/spirals Perﬂuoro polymers/spirals

100

Polysulfone/ﬁne ﬁber Polyaramide/ﬁne ﬁber

75

Polyimide/ﬁne ﬁber Silicone rubber/spirals

30

Silicone rubber/ plate-and-frame

Many

20

layer thickness determines membrane permeance.2) When the selective layer is made from a low-permeability material, such as an amorphous glassy polymer, the selective layer is made as thin as possible, typically between 0.2 to 1.0 micrometers. When the selective layer is made from a high-permeability material, such as a rubbery polymer, the selective layer has a thickness of 0.5–5.0 micrometers. Thinner rubbery membranes could be made, but the formation of concentration gradients at the membrane surface puts a limit on the maximum permeance that can be used practically [4, 5]. 8.2.1 Membrane Types and Module Configurations

Membranes can be made as ﬂat sheets in long rolls or in the form of thin, hollow tubes. Production of current gas-separation membrane modules is divided approxi2) Permeance is permeability divided by the thickness of the membrane material. Permeance is expressed in terms of gas permeation units (gpu), deﬁned as 1 1 106 cm3 (STP)/cm2 s cmHg.

8.2 Current Membrane Gas-Separation Technology

mately evenly between those based on hollow ﬁne ﬁber membranes (hydrogen separation, carbon dioxide from natural gas), capillary membranes (nitrogen from air) and ﬂat-sheet membranes (carbon dioxide from natural gas, hydrocarbon/ nitrogen, methane separations). 8.2.1.1 Hollow Fine Fiber Membranes and Modules Hollow ﬁne ﬁber membranes are extremely ﬁne polymeric tubes 50–200 micrometers in diameter. The selective layer is on the outside surface of the ﬁbers, facing the high-pressure gas. A hollow-ﬁber membrane module will normally contain tens of thousands of parallel ﬁbers potted at both ends in epoxy tube sheets. Depending on the module design, both tube sheets can be open, or as shown in Figure 8.1, one ﬁber end can be blocked and one open. The high-pressure feed gas ﬂows past the membrane surface. A portion of the feed gas permeates the membrane and enters the bore of the ﬁber and is removed from the open end of the tube sheet. Fiber diameters are small because the ﬁbers must support very large pressure differences feed-to-permeate (shell-to-bore).

Figure 8.1 The main gas-separation membrane module types [6].

j169

j 8 Membrane Gas-Separation: Applications

170

8.2.1.2 Capillary Fiber Membranes and Modules Capillary ﬁbers are produced using similar equipment to hollow ﬁne ﬁbers, but have a larger diameter, typically 200–400 micrometers, and the selective layer is formed on the inside surface of the ﬁbers. In a capillary ﬁber module, the feed gas ﬂows through the bore of the ﬁbers, as shown in Figure 8.1. A portion of the feed permeates the membrane and passes to the shell side of the modules, and is removed as permeate. The pressure difference feed-to-permeate (bore-to-shell) that capillary ﬁbers can support is limited and typically does not exceed 10–15 bar (1 bar ¼ 100 kPa). Higher pressures may rupture ﬁbers, and even a single defective ﬁber can seriously degrade the separation capability of the module. Capillary membrane modules are not as inexpensive or compact as hollow ﬁne ﬁber modules, but are still very economical. Their principal drawback is the limited pressure differential the ﬁbers can support, typically not more than 10 to 15 bar. This limitation means capillary modules cannot be used at the high pressures needed for hydrogen or natural-gas processing applications. However, capillary modules are ideally suited to lower-pressure separations, such as nitrogen from air or air dehydration. In these applications, capillary modules have essentially the entire market. 8.2.1.3 Flat-Sheet Membranes and Spiral-Wound Modules Flat-sheet membranes are made in continuous rolls 500–5000 m long. Sheets of membrane 1–2 m long are cut and folded and then packaged as spiral-wound module envelopes. A single module may contain as many as thirty membrane envelopes. Currently, the industry standard spiral-wound module is 8 inches (1 inch ¼ 2.54 cm) in diameter and about 35–40 inches long; it contains 20–40 m2 of membrane. Each membrane/module type has advantages and disadvantages [2, 7]. Hollow ﬁne ﬁbers are generally the cheapest on a per-square-meter basis, but it is harder to make very thin selective membrane layers in hollow-ﬁber form than in ﬂat-sheet form. This means the permeances of hollow ﬁbers are usually lower than ﬂat-sheet membranes made from the same material. Also, hollow ﬁne ﬁber modules require more pretreatment of the feed to remove particulates, oil mist and other fouling components than is usually required by capillary or spiral-wound modules. These factors offset some of the cost advantage of the hollow ﬁne ﬁber design. The investment in time and equipment to develop a new membrane material in a high-performance hollow ﬁne ﬁber or capillary form is far larger than that required to develop ﬂat-sheet membranes, and many materials cannot be formed into ﬁber modules at all. For this reason, ﬂat-sheet membranes, formed into spiral-wound modules, are used in many niche applications which cannot support the development costs associated with ﬁber modules. Spiral-wound modules are also competitive in the natural-gas processing area, where their general robustness is an asset. 8.2.2 Module Size

In the early days of gas separation, the average membrane module was just a few inches in diameter. In recent years, the trend has been to obtain economies of scale by

8.3 Applications of Gas-Separation Membranes

Figure 8.2 The expanding diameter of Cynara hollow-fiber membrane modules, from the first 5-inch modules of the 1980s to the 30-inch diameter behemoths now being introduced (Photo used courtesy of NATCO Group, Inc.) [9].

developing ever larger modules. Figure 8.2 shows the evolution of cellulose triacetate ﬁne ﬁber membrane modules made by Cynara (a division of NATCO) for natural-gas treatment. Spiral-wound modules are also increasing in size, from the current 8-inch diameter module to 12-inch or larger modules. The driver for these changes is the high cost of gas-separation skids. Gas-separation systems require membrane modules contained in high-pressure, code-stamped vessels. The cost of the vessels, frames, and associated pipes and valves can be several times the cost of the membrane modules. Considerable savings are obtained by packaging larger membrane modules into fewer vessels, or housing multiple modules within a single large pressure vessel [8].

8.3 Applications of Gas-Separation Membranes 8.3.1 Nitrogen from Air

The largest current application of gas-separation membranes is separation of nitrogen (N2) from air. Capillary modules formed into bore-side feed modules are used almost exclusively in this application [10, 11]. The feed air is compressed to 6–10 bar and pumped through the membrane capillaries. Oxygen (O2) permeates the membrane preferentially, leaving an oxygen-depleted, nitrogen-rich residue stream. The ﬁrst membranes used for this application were based on poly(4-methyl-1-pentene) and ethyl cellulose, and had O2/N2 selectivities of about 4. Because of the modest

j171

j 8 Membrane Gas-Separation: Applications

172

Figure 8.3 Oxygen/nitrogen selectivity as a function of oxygen permeability. The upper-bound line represents the point above which no better membranes are known [12]. This line shows the trade-off relationship between membrane permeability and selectivity.

selectivity, a signiﬁcant fraction of the nitrogen in the feed air was lost with the oxygen permeate. Within a few years, improved materials with O2/N2 selectivities of 6 to 8 were introduced. Units incorporating these membranes recovered a much higher fraction of the feed-air nitrogen. Very little change in membrane materials has taken place in the last 15 years, despite ongoing research manifested by the ﬂood of publications describing materials with improved properties. The problem is the ﬂux/selectivity trade-off relationship, illustrated by the Robeson plot. The Robeson plot shown in Figure 8.3 was created in 1991 [12]. The plot shows the O2/N2 selectivity and oxygen permeability of every membrane material reported at that time. Since 1991, other materials have been reported, but the position of the upper bound line has not moved signiﬁcantly. Many high-selectivity materials are known, but higher selectivity is always obtained at the expense of an exponential reduction in membrane permeability. Using a high-selectivity membrane means that a better separation is obtained, and so the size of the compressor required to produce a unit of product nitrogen decreases. However, this decrease in the cost of the compressor is offset by an increase in the cost of the extra membrane area needed because of the lower membrane permeability. Figure 8.4 illustrates the trade-off between compressor horsepower and membrane area for various membrane units producing the same 100 standard cubic feet per minute (scfm) (1 scf ¼ 0.0286 N m3) of 99% nitrogen. The base case is taken to be a membrane with an O2/N2 selectivity of 6 and an oxygen permeance of 8 gpu (a permeability of 0.8 Barrer and a membrane thickness of 0.1 micrometer). There is a signiﬁcant decrease in compressor horsepower as the membrane selectivity changes from 4 to 6, but thereafter, the improvement is small. However, the membrane area required to produce the same amount of

8.3 Applications of Gas-Separation Membranes

Figure 8.4 The compression power used and membrane area required for nitrogen membrane production as a function of membrane selectivity. The membrane permeability used for each selectivity is taken from the Robeson upper-bound trade-off line shown in Figure 8.3. All numbers are shown relative to a membrane with a selectivity of 6 and an oxygen permeability of 0.8 Barrer.

product nitrogen increases sharply from a selectivity of 4 to 6 and even more sharply at selectivities above 6. Barring an unexpected breakthrough, todays membranes with a selectivity of 6 to 8 are likely to continue as the industry standard. 8.3.2 Air Drying

Capillary membrane modules very similar to those used for nitrogen production are also used to produce dry air. The water molecule is smaller and more condensable than oxygen and nitrogen, so many membrane materials are available with water/air selectivities of several hundred. In air-drying applications, it is important to operate the modules in a counterﬂow mode, usually with a small sweep ﬂow from the residue gas. Some calculations illustrating the importance of counterﬂow and counterﬂow/sweep operation are shown in Figure 8.5.

j173

j 8 Membrane Gas-Separation: Applications

174

Figure 8.5 Comparison of (a) crossflow, (b) counterflow and (c) counterflow sweep module performance for the separation of water vapor from air. Membrane water/air selectivity ¼ 100, water permeance ¼ 1000 gpu.

In the crossﬂow module illustrated in Figure 8.5(a), the pooled permeate stream has a water concentration of 1.88%. The counterﬂow module illustrated in Figure 8.5(b) performs substantially better, providing a pooled permeate stream with a concentration of 3.49%. Not only does the counterﬂow module perform the separation twice as well, it also requires only about half the membrane area. This improvement is achieved because the gas permeating the membrane at the residue end of the module contains much less water than the gas permeating the membrane at the feed end of the module. Permeate counterﬂow dilutes the permeate gas at the feed end of the module with low-concentration permeate gas from the residue end of the module. This increases the water concentration driving force across the membrane and so increases the water ﬂux. In the case of the counterﬂow/sweep membrane module illustrated in Figure 8.5(c), a portion of the dried residue gas stream is expanded across a valve and used as the permeate-side sweep gas. The separation obtained depends on how much gas is used

8.3 Applications of Gas-Separation Membranes

as a sweep. In the calculation illustrated, 10% of the residue gas is used as a sweep, and the result is dramatic. The concentration of water vapor in the permeate gas is 3.05%, almost the same as for the counterﬂow module shown in Figure 8.5(b), but the membrane area required to perform the separation is one-third of the counterﬂow case. Mixing residue gas with the permeate gas improves the separation! The cause of this paradoxical result is discussed in a number of papers by Cussler et al. [13], and is illustrated graphically in Figure 8.6. Figure 8.6(a) shows the concentration of water vapor on the feed and permeate sides of the membrane module in the case of a simple counterﬂow module. On the high-pressure side of the module, the water-vapor concentration in the feed gas drops from 5000 ppm to about 1500 ppm halfway through the module and to 500 ppm at the residue end. The graph directly below the module drawing shows the theoretical maximum concentration of water vapor on the permeate side of the membrane. The actual calculated permeate-side concentration is also shown. The difference between these two lines is a measure of the driving force for water-vapor transport across the membrane. At the feed end of the module, this difference is about 15 000 ppm, but at the permeate end the difference is only about 500 ppm. Figure 8.6(b) shows an equivalent ﬁgure for a counterﬂow module in which 10% of the residue gas containing 500 ppm water vapor at 10 bar is expanded to 1 bar and introduced as a sweep gas. The water-vapor concentration in the permeate gas at the end of the membrane then falls from 4500 to 500 ppm, producing a dramatic increase in water-vapor permeation through the membrane at the residue end of the module. The result is a two-thirds reduction in the size of the module required for the separation. Counterﬂow modules are always more efﬁcient than crossﬂow modules, but the advantage is most noticeable when the membrane selectivity is much higher than the pressure ratio across the membrane and a signiﬁcant fraction of the most permeable component is being removed from the feed gas. This is the case for airdehydration membrane modules, so counterﬂow capillary modules are almost always used. With most other gas-separation applications, the advantage offered by counterﬂow designs does not offset the extra cost of making the counterﬂow type of module, so they are not widely used. 8.3.3 Hydrogen Separation

Hydrogen (H2) is a highly permeable gas; several glassy polymeric materials are known with good hydrogen permeabilities and H2/CH4 and H2/N2 selectivities of more than 50. In early applications, membranes made from these materials were used to recover hydrogen from various reactor purge streams [14]. Two typical processes are shown in Figure 8.7. The ﬁrst involves the separation of hydrogen from nitrogen, methane, and argon. In ammonia reactors, nitrogen from air and hydrogen from a methane reformer are reacted at high pressure to produce ammonia. The ammonia product is removed by cooling and condensation, leaving unreacted gas that is recycled to the reactor. Methane and argon that enter the reactor with the feed streams build up in this reactor loop, gradually degrading the performance of

j175

j 8 Membrane Gas-Separation: Applications

176

Figure 8.6 The effect of a small permeate-side, counterflow sweep on the water-vapor concentration on the permeate side of a membrane. In this example calculation, use of a sweep reduces the membrane area by two-thirds [6].

8.3 Applications of Gas-Separation Membranes

Figure 8.7 Membrane systems to recover and recycle hydrogen lost with the reactor inert-gas purge stream [6].

the reactor. To control the concentration of these inerts, a portion of the recycle loop is purged. About 4 moles of hydrogen are lost with every mole of inert gas purged. Recovery of this hydrogen is well worthwhile and easily accomplished by installing a membrane unit on the purge gas. Ninety per cent hydrogen recovery is usually obtained. Recovery of hydrogen from the recycle streams of reﬁnery hydrocrackers is a similar application [15, 16]. In these units, heavy oil is treated with hydrogen to crack C8–C12 hydrocarbons into C4–C6 molecules, and an inert-gas purge is used to remove

j177

j 8 Membrane Gas-Separation: Applications

178

Figure 8.8 Photograph of a Permea hydrogen recovery unit installed at an ammonia plant. The hollow fine fiber modules are mounted vertically [6].

methane, ethane, and propane produced as by-products. A hydrogen-permeable membrane is used to recover the hydrogen content of this purge gas. Again, 90% hydrogen recovery is obtained. The competitive technologies for these separations are cryogenic condensation and fractionation, or pressure-swing adsorption. The gas ﬂows are usually too small to make cryogenic technology applicable and the pressures involved are above the normal operating range of pressure-swing adsorption. These reasons, together with the simple ﬂow scheme, easy operation, and relatively small footprint of membrane units have made them the standard technology in these processes. Most plants use hollow ﬁne ﬁber membrane modules from Permea, Ube or Medal. A photograph of a Permea unit installed at an ammonia plant is shown in Figure 8.8. Because the gas being treated is hydrogen at high pressure, thick-walled vessels and special metallurgy are required, together with expensive controls and valves. The cost of these components far exceeds the cost of the membrane modules. 8.3.4 Natural-Gas Treatment

Removal of impurities from natural gas is, by volume of gas to be treated, the largest gas-separation application [1, 17]. About 150 trillion scf of natural gas are produced each year worldwide. All of this gas requires some treatment before it can be used. So far, membranes have captured only 5% of this market, but the membrane share is growing; currently, this is the fastest growing segment of the membrane

8.3 Applications of Gas-Separation Membranes Table 8.2 Composition specifications for natural gas delivery to the US national pipeline grid [17].

Range in US well compositions

Component

Specification

CO2

<2%

% of total US gas

Component content

72% 18% 7% 3%

<1% 1–3% 3–10% >10%

100% Water

<120 ppm

H2S

<4 ppm

— 76% 11% 4% 8%

800–1200 ppm <4 ppm 4–1000 ppm 1000–10000 ppm >10000 ppm

100% C3 þ Content Total Inert Gases (N2, He)

950–1050 Btu/scf; dew point: <20 C <4%

— 14% 86%

— 4% <4%

gas-separation industry. Raw natural gas varies substantially in composition from source to source [18]. Methane is always the major component, typically 75–90% of the total, but natural gas also contains signiﬁcant amounts of ethane, some propane and butane, and 1–3% of other higher hydrocarbons. In addition, the gas contains undesirable impurities: water, carbon dioxide, nitrogen, and hydrogen sulﬁde. Although the composition of raw gas varies widely, the composition of gas delivered to commercial pipeline grids is tightly controlled. Typical US natural-gas pipeline speciﬁcations are shown in Table 8.2. 8.3.4.1 Carbon-Dioxide Separation At present, the largest membrane application in natural-gas processing is carbon dioxide (CO2) removal. The traditional carbon-dioxide removal technology is amine absorption. Amine plants are able to reduce the carbon-dioxide concentration to less than 1%. Generally, 2–4% of the gas processed is used as fuel for the amine plant or is lost with the removed carbon dioxide. Amine plants are relatively large, complex operations with an absorber and stripping tower and the need to heat and cool large volumes of recirculating ﬂuid. Corrosion caused by amine-degradation products is a critical maintenance issue, and careful, well-monitored operating procedures are required to control the amine chemistry. Membrane plants require signiﬁcantly less operator attention and smaller units often operate unattended. For these reasons,

j179

j 8 Membrane Gas-Separation: Applications

180

Figure 8.9 Block diagram of a two-stage membrane system to process 100 million scfd of natural gas. Reproduced with permission from Ind. Eng. Chem. Res. 2008, 47(7), 2109–2121. Copyright 2008 American Chemical Society [17].

they are favored for use in remote locations, especially on offshore platforms where their lower weight and smaller footprint are additional attractive features. Both hollow ﬁne ﬁber membrane modules (cellulose triacetate from Cynara) and spiralwound membrane modules (cellulose acetate from Grace and Separex, perﬂuoro membranes from MTR) are used. A block diagram of a typical 100 million scfd gas processing plant is shown in Figure 8.9. A plant of this size costs about US$20 million, depending on location and overall complexity. More than 30 plants of this size or larger have been installed. A photograph of the Kandanwari, Pakistan, plant installed by UOP is shown in Figure 8.10. This plant treats 500 million scfd of gas and is currently

Figure 8.10 Photograph of the 500-million scfd CO2 removal plant installed by UOP at Kandanwari, Pakistan. The UOP Separex system reduces CO2 content of a natural-gas stream from 6.5 to 2% CO2. Photo used courtesy of UOP, LLC.

8.3 Applications of Gas-Separation Membranes

Figure 8.11 Robeson plot of CO2/CH4 selectivity versus membrane permeability and permeance [12]. The points shown are based on low-pressure, pure-gas measurements. The performance of commercial membranes when used to separate carbon dioxide from high-pressure natural gas is shown on the same figure for comparison.

the worlds largest membrane gas-separation unit. Even larger plants are on the drawing board. The Kandanwari plant and almost all of the large carbon-dioxide separation plants installed to date have used cellulose acetate or triacetate membranes with a CO2/CH4 selectivity in operation of 10 to 20. This fact may seem strange, since membrane materials with selectivities of 50 or more and higher permeabilities than cellulose acetate are routinely reported. Figure 8.11 shows a Robeson plot for carbon-dioxide/methane separations. The position of todays celluloseacetate membranes is shown on the plot. The commercial membranes in use have half the reported selectivity of the best upper-bound materials. This difference reﬂects the difference between selectivity estimated from the ratio of pure methane and carbon-dioxide permeability measurements and the real world selectivity measured with high-pressure gas mixtures containing plasticizing components, including not only carbon dioxide, but also water, heavy hydrocarbons, and aromatics [19, 20]. Developing membranes and processes that are able to operate under real-world conditions is where the bulk of industry research is focused.

j181

j 8 Membrane Gas-Separation: Applications

182

8.3.4.2 Separation of Heavy Hydrocarbons Rubbery polymers, most commonly silicone rubber, are used to separate heavy hydrocarbons from natural gas. The traditional technology for this separation is cooling and condensation, but membrane units have found a use in small applications where simplicity and ease of operation are needed. For example, at remote locations, raw untreated natural gas is used as fuel for ﬁeld compressor engines or power-generating turbines. This gas is often produced in association with oil, so it can contain high levels of heavy hydrocarbons and aromatics that cause coking and preignition when the gas is used as engine fuel. The design of a simple membrane unit to treat such a gas by preferentially permeating the heavy components is shown in Figure 8.12. The feed to the unit is a slip stream from the compressed pipeline gas. The clean residue, stripped of the heavy hydrocarbons, is used as compressor engine fuel, and the heavy hydrocarbons are recycled to the suction side of the ﬁeld compressor [21]. 8.3.4.3 Nitrogen Separation from High-Nitrogen Gas A second application of rubbery membranes in natural-gas processing is separation of nitrogen from high-nitrogen gas. Pipeline gas must normally contain less than 4% nitrogen, but the pipeline operator will often accept high-nitrogen gas if sufﬁcient low-nitrogen gas is available to dilute the off-spec gas. When dilution is not possible, cryogenic, adsorption, or membrane treatment of the gas is required [22–24]. Methane is about three times more permeable than nitrogen through silicone rubber membranes, so these membranes can be used to perform a separation. Because the membrane selectivity is low, multistep or multistage systems must be used. The design of a two-step nitrogen separation plant installed in a Sacramento River Delta gas ﬁeld in California is shown in Figure 8.13 [17]. The feed gas contains 16% nitrogen. The heating value of the gas is 900 Btu/scf (1 Btu ¼ 1.0550 103 joules). The pipeline accepts gas for dilution with lownitrogen gas if the heating value is raised to 990 Btu and the nitrogen content reduced to about 9%. To reach this target, the feed gas, at a pressure of 980 psia, is passed through three sets of modules in series. The permeate from the front set of modules is preferentially enriched in methane, ethane, and the C3 þ hydrocarbons, and the nitrogen content is reduced to 9% nitrogen. These changes raise the heating value of the gas to 990 Btu/scf. This gas is compressed and sent to the pipeline. The residue gas (containing 22% nitrogen) is sent to a second membrane step where it is concentrated to 60% nitrogen. The permeate from the second step contains 18% nitrogen and is recycled to mix with the feed gas. The residue gas from the second unit is then sent to a ﬁnal small module unit to be fractionated. The permeate gas from the ﬁnal unit contains 40% nitrogen and is used as fuel for the compressor engines. The ﬁnal residue contains 65–70% nitrogen, and is essentially stripped of all C3 þ hydrocarbons. This gas is vented. The unit recovers about 96% of the heating value of the feed gas in the product stream, about 2% of the gas is used as compressor fuel and another 2% is lost with the nitrogen vent.

8.3 Applications of Gas-Separation Membranes

Figure 8.12 Block diagram and photograph of a membrane fuel-gas conditioning unit (FGCU) used for a field gas compressor engine (the unit uses silicone rubber membranes in spiral-wound modules). The membrane modules are

contained in the horizontal pressure vessels. The unit produces 0.5–1.0 MMscfd of clean gas. Reproduced with permission from Ind. Eng. Chem. Res. 2008, 47(7), 2109–2121. Copyright 2008 American Chemical Society [17].

8.3.5 Vapor/Gas Separations in Petrochemical Operations

In the separation of vapor/gas mixtures, rubbery polymers, such as silicone rubber, can be used to permeate the more condensable vapor components, or glassy polymers can be used to permeate the smaller gases. Although glassy, gas-permeable membranes have been proposed for a few applications, most installed plants use

j183

j 8 Membrane Gas-Separation: Applications

184

Figure 8.13 Flow diagram and photograph of a 12 MMscfd membrane nitrogen removal plant installed on a high-nitrogen gas well in the Sacramento River Delta region of California. Reproduced with permission from Ind. Eng. Chem. Res. 2008, 47 (7), 2109–2121. Copyright 2008 American Chemical Society [17].

rubbery vapor-permeable membranes, often in conjunction with a second process such as condensation [25–27]. The ﬁrst plants were used in the early 1990s to treat gasoline terminal vent gases or chloroﬂuorocarbon (CFC) vapor vents from industrial refrigeration plants. Membranes are now used to recover hydrocarbons and processing solvents from petrochemical plant purge streams. Some of these streams are large, and discharge vapors with a recovery value of US$2–5 million/y. One of the most successful petrochemical applications is illustrated in Figure 8.14: treatment of resin-degassing vent gas in a polyoleﬁn plant [28]. In these plants, oleﬁn monomer, catalyst, solvents, and other coreactants are fed at high pressure into a polymerization reactor. The polymer product (resin) is removed from the reactor and separated from excess monomer in a ﬂash-separation step. The recovered monomer is recycled to the reactor. Residual monomer is removed from the resin powder by

8.3 Applications of Gas-Separation Membranes

Figure 8.14 Flow diagram showing the use of hydrocarbonpermeable membranes to recover unreacted monomers from a polyolefin plant resin degassing unit. The photograph is of a system installed by MTR in Qatar in 2007.

stripping with nitrogen in a ﬂuidized bed. The composition of the resulting degassing vent stream varies greatly, but it usually contains 20–30% mixed hydrocarbon monomers in nitrogen. The monomer content of the gas represents about 1% of the hydrocarbon feedstock entering the plant. This amount might seem small, but because polyoleﬁn plants are so large, the recovery value of the stream can be signiﬁcant. About 40 such plants are in use worldwide. Degassing a vent stream with a membrane system is shown in Figure 8.14. The compressed vent gas is sent to a condenser, where a portion of the hydrocarbon vapors is removed as a liquid. The uncondensed hydrocarbons and nitrogen are separated in the membrane unit, which produces a hydrocarbon-enriched permeate and a puriﬁed nitrogen stream (>98% nitrogen). The nitrogen stream is recycled to the resin degasser. The hydrocarbon-enriched permeate is returned to the front of the compressor for hydrocarbon recovery; the hydrocarbon liquid stream from the condenser is upgraded in the monomer puriﬁcation section of the plant and then

j185

j 8 Membrane Gas-Separation: Applications

186

Figure 8.15 Flow schematic of the OPW Vaporsaver unit to minimize gasoline vapor emissions. More than 1000 of these systems have been installed at United States retail gasoline stations.

recycled to the reactor. Similar compression–condensation-membrane separation systems have been installed to separate hydrocarbon/nitrogen mixtures produced by a wide variety of petrochemical processes. More than 20 large systems have also been installed at gasoline terminals to separate gasoline vapors from vent streams produced during gasoline loading operations [29]. At the other end of the scale, more than 1000 small systems have been installed at gasoline stations to minimize the release of hydrocarbon vapors to the atmosphere. These systems use a small pump to draw air and vapors from the gasoline dispensing nozzle (see Figure 8.15). For every liter of gasoline dispensed from the pump, as much as two liters of air and gasoline vapor are returned to the storage tank. Build-up of air in the tank leads to atmospheric releases of gasoline vapor-laden air from the tank head space. Systems ﬁtted with membranes to recover gasoline vapors and return them to the storage tank reduce hydrocarbon emission by 95–99%.

8.4 Future Applications

The applications described above cover the bulk of the current industrial membrane gas-separation business. A number of applications in various stages of development that could become commercial in the next few years are described brieﬂy below. 8.4.1 CO2/N2 Separations

Worldwide, approximately 5000 coal-based electric power plants release a total of 10 billion metric tons of carbon dioxide into the atmosphere each year. Separation of this carbon dioxide from power-plant ﬂue gas and sequestration as liquid carbon dioxide

8.4 Future Applications

Figure 8.16 Block diagrams illustrating the use of CO2-selective membranes to separate CO2 from power-plant flue gas. These calculations are based on a power plant producing 600 MWe of power. The membrane permeances and size of the membrane units are based on membranes with a CO2 permeance of 1000 gpu and a CO2/N2 selectivity of 40.

into deep aquifers/salt domes is a target of research programs around the world [30]. The cost of carbon-dioxide capture from ﬂue gas depends on the type of power plant producing the gas, the fuel input source (coal, oil, or natural gas), and the capture technology (absorption, adsorption, chemical scrubbing, or membranes) [30, 31]. Currently, carbon-dioxide capture with amine absorption seems to be the leading candidate technology – although membrane processes have been suggested [32]. The use of selective membranes to separate carbon dioxide from ﬂue gas is illustrated in Figure 8.16. Figure 8.16(a) shows a simpliﬁed ﬂow diagram of a conventional power plant. For ease of calculation, the fuel input is assumed to be 150 tons/h of carbon as medium-volatility coal. Combustion of this amount of fuel with an excess of air would generate 2.26 106 m3/h of ﬂue gas containing 13% carbon dioxide. This hypothetical plant would produce approximately 600 MWe of electric power (at 10 000 Btu heat/kW power). Figure 8.16(b) shows a single-stage membrane process for treating the ﬂue gas. The process cuts carbon dioxide emissions by 90%. In this process, the ﬂue gas is compressed and cooled, which removes most of the water vapor. The gas, which contains about 13% carbon dioxide, then passes across the surface of a CO2permeable membrane, producing a permeate containing 38% carbon dioxide and a pressurized residue stream containing 2.1% carbon dioxide. The residue stream is expanded through the turbine compressor, which reduces the power consumption of this unit by more than 50%. The net energy consumption of the turbine compressor is about 110 MWe, or 19% of the electric power produced. A further 15% is required to concentrate, compress, and condense the carbon dioxide in the low-pressure

j187

j 8 Membrane Gas-Separation: Applications

188

permeate stream, to produce pure, high-pressure, supercritical carbon dioxide for pipeline sequestration. So approximately 35% of the power plants electricity is used to separate and sequester the carbon dioxide produced. The membrane plant is also very large. The design shown in Figure 8.16 uses membranes with very high permeance and selectivity, but still requires 600 000 m2 of membrane. Very low cost membranes and membrane modules are needed to make this process viable. Innovative process designs are being developed to reduce the size of the membrane unit and the energy needed to separate, condense and inject the carbon dioxide. It seems possible to reduce the energy consumption of the membrane process to about 20–25% of the power plant output. If this work is successful and these membrane plants are built, this application will dwarf all other gas-separation membrane processes. 8.4.2 CO2/H2 Separations

The production of hydrogen from coal is expected to become an important aspect of future energy supply [33]. A block ﬂow diagram of one of the proposed coal-tohydrogen production processes is shown in Figure 8.17. The process starts with a gasiﬁcation step, in which coal is reacted with oxygen and steam at high temperature and pressure to produce hot syngas, containing mainly hydrogen and carbon monoxide. This gas is scrubbed of tars, particulates and sulfur compounds and cooled to 400–450 C. The gas then passes through two catalytic shift reactors, at 400 and 250 C, respectively, to convert carbon monoxide to carbon dioxide and hydrogen via the reaction (8.1) ð8:1Þ CO þ H2 O ! CO2 þ H2

Figure 8.17 Block diagram of a coal-to-hydrogen production plant. Many variants of the process have been proposed.

8.4 Future Applications

The gas is then cooled to 30–50 C and the carbon dioxide is removed by amine absorption or other processes. The remaining impurities – carbon monoxide, methane, nitrogen, argon – are removed in a ﬁnal pressure-swing adsorption (PSA) step to yield >99.5% pure hydrogen. One of the main problems with this process is that the carbon dioxide is removed by the amine unit as a low-pressure gas. This gas must be compressed to 80 bar to be pipelined for sequestration. This compression step alone requires massive compressors and uses 4–5% of the total power output of the plant. The amine treatment step itself uses even more energy, so the total energy consumption is 15% of the power produced by the plant. One possible improvement suggested by many authors is the use of hydrogenpermeable membranes instead of an amine unit to remove carbon dioxide [30]. These membranes separate the hydrogen as a low-pressure permeate gas, and the carbon dioxide is left as a high-pressure residue gas. Condensation of the carbon dioxide to liquid carbon dioxide is then much less costly. A number of zeolite [34], ceramic and metal membranes [35, 36] have the required selectivity, but are expensive to produce, so the plants capital cost may be high. Some polymerics have selectivities in the range of 5–15 and could also be used. Unfortunately, permeabilities are relatively low [37, 38]. An alternative approach is to use membranes that preferentially permeate the carbon dioxide and retain the hydrogen. A number of polar polyethylene oxiderelated polymers [39, 40] have been found with CO2/H2 selectivities of up to 10 and good permeabilities. Development of this type of membrane has promise. 8.4.3 Water/Ethanol Separations

A pervaporation plant for the dehydration of bioethanol was in operation in 1982. In the succeeding two decades, many small systems were installed, but the technology never really got off the ground [41]. A number of developments are about to change this. First, since about 2003, a very large bioethanol industry has grown up in the United States and Brazil. About 7 billion gallons (1 ﬂuid gal ¼ 3.785 liters) of ethanol were produced from corn in the USA in 2007, and a further 6 billion gallons were produced from sugar in Brazil. Worldwide production in 2007 was about 14 billion gallons. Production may reach 30 billion gallons by 2012, especially if cellulose-toethanol technology (now at the demonstration stage) becomes commercial. The current ethanol dehydration technology – two-stage distillation followed by a molecular-sieve dryer, as shown in Figure 8.18(a) – uses approximately 16 000–20 000 Btu of energy/gal of ethanol produced. This is about 20% of the energy value of the ethanol produced. There is a considerable interest in membrane technology that would be lower in cost and less energy intensive. Most of the current industrial development efforts are focused on processes that separate water from the overhead ethanol/water vapor of the distillation column, replacing the molecular sieve drier as shown in Figure 8.18(b). The overhead vapor mixture is sent to a water-permeable membrane, producing a dry ethanol residue and a low-pressure permeate enriched in water, which is recycled to the column. Another option, shown in Figure 8.18(c), is to use the membrane-separation step to replace

j189

j 8 Membrane Gas-Separation: Applications

190

Figure 8.18 Flow schemes of the separation train of a 50-million gallon/y bioethanol plant. Current technology is illustrated in (a). Pervaporation membranes can be used to replace the molecular-sieve drier of the plant (b) or vapor-permeation membranes can be used to replace the rectifier column and molecular-sieve units (c).

8.5 Summary/Conclusion

both the second distillation column and the molecular-sieve drier. Depending on the operating conditions of the distillation column, the membranes for these options must separate ethanol/water liquid or vapor mixtures at temperatures of 100–120 C. Ceramic membranes can perform this separation, but are likely to be expensive; polymeric membranes cost much less, but are often unstable in the presence of hot ethanol/water mixtures [42]. The academic literature is ﬂooded with papers describing the separation of ethanol-water mixtures at 50–60 C. Research of this type has no industrial relevance for this application. 8.4.4 Separation of Organic Vapor Mixtures

The only vapor/vapor mixture currently separated on an industrial scale is ethanol/ water. These components have sufﬁciently different properties that membranes can be found that retain good selectivities, even at high temperatures. Finding membranes that can separate hydrocarbon mixtures has proven much more difﬁcult. The most widely studied vapor mixture is propylene/propane. In the United States, over 34 billion pounds (1 billion pounds ¼ 0.446 million metric tons) of propylene were produced in 2004. All of this propylene had to be separated from propane by distillation. These compounds have very similar boiling points, and the separation requires a very pure propylene overhead (99.5 wt%) and a relatively pure (95 wt%) propane bottoms stream. This means very large columns with 180–240 trays are needed. The distillation column reﬂux ratios are also as high as 15 to 35, which leads to very high energy costs. The problem with use of polymeric membranes in this application is plasticization, leading to much lower selectivities with gas mixtures than the simple ratio of pure-gas permeabilities would suggest. For this type of separation, a Robeson plot based on the ratio of pure-gas permeabilities has no predictive value. Although membranes with pure-gas propylene/propane selectivities of 20 or more have been reported [43, 44], only a handful of membranes have been able to achieve selectivities of 5 to 10 under realistic operating conditions, and these membranes have low permeances of 10 gpu or less for the fast component (propylene). This may be one of the few gas-separation applications where ceramic or carbon membranes have an industrial future.

8.5 Summary/Conclusion

The membrane gas-separation industry has come a long way from its starting point in 1980. Plant operators at that time were only ready to consider a membrane solution to their problems if the economics were too good to ignore and if conventional technology could not do the job. Today, membrane gas-separation plants are commonplace and are recognized as being reliable, efﬁcient, and cost effective. As a consequence, the industry is growing quickly and prospects for future growth look good as membranes compete successfully against absorption, adsorption, and

j191

j 8 Membrane Gas-Separation: Applications

192

cryogenic technology. However, the real future for membranes will be in new applications. If CO2 sequestration is ever used, it will require many very large plants separating CO2/H2 and CO2/N2 mixtures; these are great applications for membrane separations. Separation of water from ethanol is also likely to be adopted by industry in the next few years. And ﬁnally, there is the separation of propylene/propane mixtures. Ten years ago, I thought development of membranes to separate these mixtures was just around the corner. I was wrong. Useful membranes for this separation remain an unsolved problem, but hope springs eternal, and several laboratories are still pursuing projects to make these membranes. The ﬁrst thirty years have set the stage and have overcome resistance to the use of membranes in gas separation; the next thirty years will certainly extend their use to larger and more varied applications.

References 1 Spillman, R.W. (1989) Economics of gas separation by membranes. Chemical Engineering Progress, 85, 41. 2 Koros, W.J. and Fleming, G.K. (1993) Membrane-based gas separation. Journal of Membrane Science, 83, 1. 3 Baker, R.W. (2002) Future directions of membrane gas-separation technology. Industrial & Engineering Chemistry Research, 41, 1393. 4 L€ udtke, O., Behling, R.-D. and Ohlrogge, K. (1998) Concentration polarization in gas permeation. Journal of Membrane Science, 146, 145–157. 5 He, G., Mi, Y., Yue, P.L. and Chen, G. (1999) Theoretical study on concentration polarization in gasseparation membrane processes. Journal of Membrane Science, 153, 243–258. 6 Baker, R.W. (2004) Membrane Technology and Applications, John Wiley & Sons Ltd, Chichester, UK. 7 Zolandz, R.R. and Fleming, G.K. (1992) Design of Gas Permeation Systems, in Membrane Handbook (eds W.S. Ho and K.K. Sirkar), Van Nostrand Reinhold, New York. 8 Kaschemekat, J., Fulton, D. and Wynn, N. (2008) Gas-separation Membrane Module Assembly, U.S. Patent 7,404,843.

9 Blizzard, G., Parro, D. and Hornback, K. (2005) CO2 separation membranes a critical part of the mallat CO2 removal facility. Proceedings of the Laurance Reid Gas Conditioning Conference, University of Oklahoma. 10 Prasad, R., Shaner, R.L. and Doshi, K.J. (1994) Comparison of Membranes with Other Gas-separation technologies, in Polymeric Gas-separation membranes (eds D.R. Paul and Y.P. Yampolskii), CRC Press, Boca Raton, FL, pp. 531–614. 11 Prasad, R., Notaro, F. and Thompson, D.R. (1994) Evolution of membranes in commercial air separation. Journal of Membrane Science, 94, 225. 12 Robeson, L.M. (1991) Correlation of separation factor versus permeability for polymeric membranes. Journal of Membrane Science, 62, 165. 13 Wang, K.L., McCray, S.H., Newbold, D.N. and Cussler, E.L. (1992) Hollow ﬁber air drying. Journal of Membrane Science, 72, 231. 14 Henis, J.M.S. (1994) Commercial and Practical Aspects of Gas-separation membranes, in Polymeric Gas-separation membranes (eds D.R. Paul and Y.P. Yampolskii), CRC Press, Boca Raton, FL, pp. 441–512. 15 Bollinger, W.A., MacLean, D.L. and Narayan, R.S. (1982) Separation systems

References

16

17

18

19

20

21

22

23

24

25

for oil reﬁning and production. Chemical Engineering Progress, 78, 27. MacLean, D.L., Bollinger, W.A., King, D.E. and Narayan, R.S. (1986) Gas Separation Design with Membranes, in Recent Developments in Separation Science (eds N.N. Li and J.M. Calo), CRC Press, Boca Raton, FL, p. 9. Baker, R.W. and Lokhandwala, K. (2008) Natural-gas processing with membranes: an overview. Industrial & Engineering Chemistry Research, 47 (7), 2109–2121. Hugman, R.H., Springer, P.S. and Vidas, E.H. (1990) Chemical Composition in Discovered and Undiscovered Natural Gas in the Lower-48 United States, Report No. GRI-90/0248, Energy and Environmental Analysis, Inc., Arlington, VA. Sanders, E.S. (1988) Penetrant-induced plasticization and gas permeation in glassy polymers. Journal of Membrane Science, 37 (1), 63–80. Visser, T., Koops, G.H. and Wessling, M. (2005) On the subtle balance between competitive sorption and plasticization effects in asymmetric hollow ﬁber gasseparation membranes. Journal of Membrane Science, 252, 265–277. Jariwala, A., Lokhandwala, K. and Baker, R.W. (2006) Only raw sour gas for engine fuel? Proven membrane process cleans gas for engines. Proceedings of the Laurance Reid Gas Conditioning Conference, University of Oklahoma. Tannehill, C.C. (1999) Nitrogen Removal Requirement from Natural Gas, Topical Report, Report No. GRI-99/0080, Gas Research Institute, Chicago, IL. Mitariten, M. (2004) Economic nitrogen removal. Hydrocarbon Engineering, 9 (7), 53–57. Baker, R.W., Lokhandwala, K.A., Pinnau, I. and Segelke, S. (1997) Methane/Nitrogen Separation Process, U.S. Patent 5,669,958. Wijmans, J.G.Process for Removing Condensable Components from Gas Streams, U.S. Patents 5,199,962, 1993, and 5,089,033, 1992.

26 Baker, R.W. and Wijmans, J.G. (1994) Membrane Separation of Organic Vapors from Gas Streams, in Polymeric Gas Separation Membranes (eds D.R. Paul and Y.P. Yampolskii), CRC Press, Boca Raton, FL, pp. 353–397. 27 Ohlrogge, K., Wind, J. and Belling, R.D. (1995) Off gas puriﬁcation by means of membrane vapor separation systems. Separation Science and Technology, 30, 1625. 28 Baker, R.W. and Jacobs, M. (1996) Improved monomer recovery from polyoleﬁn resin degassing. Hydrocarbon Processing, 75(3), 49. 29 Ohlrogge, K. and St€ urken, K.S. (2001) The Separation of Organic Vapors from Gas Streams by Membranes, in Membrane Technology in the Chemical Industry (eds S.P. Nunes and K.-V. Peinemann), Wiley-VCH, Weinheim, Germany. 30 Hendriks, C. (1994) Carbon-dioxide removal from Coal-Fired Power Plants, 1st edn, Kluwer Academic Publishers, Boston, MA. 31 (2005) Carbon Capture and Sequestration Systems Analysis Guidelines, U.S. Department of Energy, National Energy Technology Laboratory. 32 Favre, E. (2007) Carbon dioxide recovery from post-combustion processes: can gas permeation membranes compete with absorption? Journal of Membrane Science, 294, 50. 33 U.S. DOE Ofﬁce of Fossil Energy (2006) Hydrogen from Coal RD&D Plan, http:// www.fossil.energy.gov/programs/fuels/ publications/programplans/2005/ Hydrogen_From_Coal_RDD_Program_ Plan_Sept.pdf. 34 Bredesen, R., Jordal, K. and Bolland, O. (2004) High-temperature membranes in power generation with CO2 capture. Journal of Membrane Science, 43, 1129–1158. 35 Ma, Y., Mardilovich, I.P. and Engwall, E.E. (2003) Thin composite palladium and palladium/alloy membranes for hydrogen separation. Annals of the New York Academy of Sciences, 984, 346–360.

j193

j 8 Membrane Gas-Separation: Applications

194

36 Dolan, M.D., Dave, N.C., Ilyushechkin, A.Y., Morpeth, L.D. and McLennan, K.G. (2006) Composition and operation of hydrogen-selective amorphous alloy membranes. Journal of Membrane Science, 285, 30–55. 37 Young, J.S., Long, G.S. and Espinoza, B. (2006) Cross-linked Polybenzimidazole Membrane for Gas Separation, U.S. Patent 6,997,971. 38 Pesiri, D.R., Jorgensen, B.J. and Dye, R.C. (2003) Thermal optimization of polybenzimidazole meniscus membranes for the separation of hydrogen, methane, and carbon dioxide. Journal of Membrane Science, 218, 11–18. 39 Lin, H. and Freeman, B.D. (2005) Materials selection guidelines for membranes that remove CO2 from gas mixtures. Journal of Molecular Structure, 739, 57–74. 40 Lin, H., Van Wagner, E., Freeman, B.D., Toy, L.G. and Gupta, R.P. (2006)

41

42

43

44

Plasticization-enhanced H2 puriﬁcation using polymeric membranes. Science, 311, 639–642. Wynn, N. (2001) Pervaporation comes of age. Chemical Engineering Progress, 97, 66. Vane, L.M. and Alvarez, F.R. (2008) Membrane-assisted vapor stripping: energy-efﬁcient hybrid distillation vapor permeation processes for alcohol-water separation. Journal of Chemical Technology and Biotechnology, 83(9), 1275–1287. Staudet-Bickel, C. and Koros, W.J. (2000) Oleﬁn/parafﬁn gas separations with 6FDA-based polyimide membranes. Journal of Membrane Science, 170, 205. Shimadzu, A., Miyazaki, T., Maeda, M. and Ikeda, K. (2000) Relationship between the chemical structure and the solubility diffusivity and permselectivity of propylene and propane in 6-FDA-based polyimides. Journal of Polymer Science Part B-Polymer Physics, 38, 2525.

j195

9 CO2 Capture with Membrane Systems Rune Bredesen, Izumi Kumakiri, and Thijs Peters

9.1 Introduction 9.1.1 CO2 and Greenhouse-Gas Problem

Economical growth and well fare are directly linked to energy consumption. The worlds energy needs are currently mainly provided by combustion of fossil fuels (85%), making CO2 the most important anthropogenic greenhouse gas (GHG) [1]. The UN Intergovernmental Panel on Climate Change (IPCC) and the International Energy Agency (IEA) have recently published reports [2, 3] that forecast a substantial increase in energy demand and GHG emissions in the coming years. A Reference Scenario used by IEA anticipates that GHG emissions increase by 57% between 2005 and 2030 [3]. Only strong political measures can stabilize the emission in one or two decades, and reduce it in longer terms, thereby limiting the expected dramatic increase in average global temperature. The GHG emissions can only be reduced by parallel actions to improve energy efﬁciency, changing to renewable and nonfossilfuel-based energy sources, and through broad deployment of carbon capture and storage (CCS) technology. CCS technology is already in use in sweetening of natural gas [2, 4], for example, the Statoil Sleipner natural-gas production installations in the North Sea capture and store nearly 1 Mtonne CO2/year [4]. CCS technology is most cost effective at large-scale point emissions, like power plants and some large industries. Currently, no full-scale power plant exists with CCS technology, but erection of several large-scale demonstration plants have been announced [5–7]. In the short-term perspective, demonstration at large scale is technologically valuable, however, for broad deployment of CCS technology the cost of CO2 capture, representing 60–80% of the total CCS cost depending on fuel and process, must become economical on commercial terms. The capture cost is therefore the most critical issue to be solved.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 9 CO

196

2

Capture with Membrane Systems

To comply with IPCC recommendations, substantial GHG emission reduction from small- and medium-scale sources is also necessary [8]. A change from distributed use of fossil fuel to large-scale production, including CO2 capture, and distribution of hydrogen and electricity is a plausible route to such reduction. A relevant question is whether capture and storage from small- and medium-scale sources would be economically and practically feasible. A recent IEA report [9] suggests that distributed collection through pipelines from medium-size sources in industrialized areas (North West of England was used as a case study) could be an alternative, assuming that safety issues are handled satisfactorily. The report concludes that emission sources of the order of 5 ktonne CO2/yr only constitutes 4% of the total emission, and that this CO2 is considerably more expensive to collect compared to the emission from large (emission >1 Mtonne CO2/yr, 73% of total) and medium-size sources (emission > 45 ktonne CO2/yr, 23% of total). The medium-size sources are typically energy-intensive industries, like iron and steel, glass, cement, pulp and paper, bulk chemicals production, and reﬁnery sites. Although this chapter mainly focuses on membranes for CO2 mitigation in power generation, we believe that future developments and implementation of cost-effective sustainable technology in other sectors will involve extensive use of energy-lean membrane technology. 9.1.2 CO2 Capture Processes and Technologies

Capture of CO2 from fossil-fuel-based large-scale power generation sites is commonly described along three processes routes (Figure 9.1(a)–(c)), (i) postcombustion capture, (ii) precombustion decarbonization, and (iii) oxy-fuel (combustion of fuel in oxygen without the presence of nitrogen). The ﬁgure does not include separation steps for gas cleaning in the three process routes (e.g., removal of sulfur, mercury, particles), as this falls outside the scope of this chapter. The key separation processes envisaged for the three capture routes, indicated as membrane separation processes, are listed in Table 9.1. (i) In postcombustion processes, CO2 is separated at ambient pressure from the exhaust gas after combustion of fossil fuel in air. Depending on the fuel and the process design, the concentration of CO2 in the exhaust gas may typically vary from 30 to 5 vol.%, with N2 being the main component. Separation is usually considered at close to ambient temperature, but it could in principle be performed at higher temperatures during cooling of the exhaust gas. (ii) In precombustion decarbonization processes, carbon is separated from the fuel before combustion. This is typically done ﬁrstly by converting the fuel to synthesis gas (CO þ H2), and secondly, by transformation of the heating value of CO to H2 by the water gas shift (WGS) reaction. The synthesis gas may be produced by gasiﬁcation of, for example, coal, or reforming of hydrocarbons. The produced CO2 and H2 are separated, a process being facilitated by the high CO2 concentration and pressure. Elimination of carbon as solid, for example, in

9.1 Introduction

Figure 9.1 Basic schematic diagram of (a) postcombustion capture, (b) precombustion decarbonization, and (c) oxy-fuel processes.

plasma or catalytic cracking of hydrocarbons to produce H2, may also be categorized as precombustion processes. (iii) In oxy-fuel processes, combustion of fossil fuel is carried out using oxygen instead of air, thus avoiding N2 dilution of the exhaust gas. To reduce the high temperature generated, CO2 from the cooled-down exhaust gas is recycled and used as a dilutant in the combustion [10]. For solid oxide fuel cells (SOFC) based on oxygen-ion-conducting electrolyte membranes, an overstoichiometric amount of air ﬂow can be used on the cathode side to control temperature. Current commercial CO2 capture methods [11] include absorption (chemical and physical), adsorption, cryogenic processes, and polymeric membrane technologies. CO2 capture technologies have not yet been integrated in any large-scale fossil-fuelbased power generation plant [12]. In the coming years, there is a deﬁnite need to establish CO2 capture experience on a large scale, to improve existing technologies and to develop alternative technologies and processes, to lower the cost and energy consumption related to CO2 capture. To illustrate the magnitude of separation cost of existing technologies, Table 9.2 gives values for 4 types of power plants. The cost of the separation units constitutes 20–40% of the total capital cost of the power plants with CO2 capture. Traditional absorbers and distillation systems are energy consuming and therefore reduce net efﬁciency of the power plant signiﬁcantly. The efﬁciency penalty is also negatively affected by the increasing price of fossil fuels, thus making energy-lean technologies more competitive. In the following

j197

H2O þ CO2

Anode exhaust gas SOFC containing unconverted fuel

H2 separation

H2

CO2 þ H2O CO þ H2 CO2 CO2 CO2 CO2 CO2 H2 H2

SOFC cathode air stream Air (compression not necessary) Flue gas N2/H2O/CO2 Flue gas H2O/CO2 Synthesis gas CO/H2/H2O/CO2 Natural gas þ steam Natural gas CH4/CO2 Synthesis gas CO/H2/H2O/CO2 Natural gas þ steam

CO2 separation

O2

Compressed air

O2 separation

Main components, permeate

Oxygen for gasiﬁcation of fossil fuels (coal, heavy oil, biomass) Oxygen for oxy-fuel combustion Oxygen for combustion of remaining fuel from various process streams Combustion of remaining fuel in SOFC anode exhaust gas Partial oxidation of natural gas in precombustion Postcombustion capture Recycling of CO2 in oxy-fuel process Precombustion decarbonization (WGS) Precombustion decarbonization (MSR þ WGS) Natural gas sweetening Precombustion decarbonization (WGS), H2 production for fuel cells Precombustion decarbonization (MSR þ WGS), H2 production for fuel cells, and so on. Combustion of SOFC anode off-gas with cathode air stream (SOFC-GT concept). Recycling of fuel or for PEM, and so on.

Relevance

2

Main components, feed

j 9 CO

Membrane process

Table 9.1 Typical (membrane) separation processes for CO2 mitigation.

198

Capture with Membrane Systems

9.2 Membrane Processes in Energy Systems with CO2 Capture Table 9.2 Capital cost and % cost of CO2 separation unit for different power plants with CO2 capture, after [13].

Plant type

Capital cost ($/kW)

NGCC (GTCC)

916

PC

1962

IGCC

1831

PC

2417

Separation unit Postcombustion Amine chemical absorption Postcombustion Amine chemical absorption Precombustion Air separation (O2 production) WGS/selexol physical absorption Oxy-fuel Air separation (O2 production) CO2 distillation

Capital cost (% total)

24 18 18 13 32 7

NGCC/GTCC, Natural gas combined cycle (often termed gas-turbine combined cycle); PC, Pulverized-coal-ﬁred power plant; IGCC, Integrated gasiﬁcation combined cycle, Oxy-fuel (PC boiler) plant. Flue-gas desulfurization and air particulate control is included in the total cost, but not in the separation unit cost.

we will ﬁrst continue looking at integration of membrane processes in energy systems with CO2 capture, and then highlight relevant membrane properties and critical issues related to this technology.

9.2 Membrane Processes in Energy Systems with CO2 Capture 9.2.1 Processes Including Oxygen-Separation Membranes

In precombustion CO2 capture schemes, oxygen-separation membranes can be used for synthesis gas production, either by partial oxidation of natural gas, or in gasiﬁcation of fossil fuel (coal, oil or biomass) [14, 15]. In partial oxidation, dense ceramic oxygen-separation membrane can provide oxygen directly to high pressure natural gas at 800–1000 C in a membrane reactor. A technical and economical evaluation [14] of reforming processes integrating oxygen-ion-conducting membranes has shown that an oxygen ﬂux exceeding 10 mL/cm2 min produces a competitive technology (Figure 9.2). Such ﬂux values have been reported for existing membranes) [16]. For IGCC coal-based processes, where pure oxygen is used for gasiﬁcation to produce synthesis gas, a signiﬁcant cost reduction may appear by changing the production method from cryogenic to membrane technology (Figure 9.3) [15]. This reduction has been estimated to be 35% in capital cost and 37% in power consumption [17], for a 438-MW IGCC plant.

j199

j 9 CO

200

2

Capture with Membrane Systems

Figure 9.2 Sensitivity of investment cost to membrane price and O2 flux in autothermal reforming, after [14]. Cost employing conventional cryogenic air separation technology shown by broken line.

Another IGCC process-simulation study, employing steam as the membrane permeate-side sweep gas, reported that the net efﬁciency of the plant is critically dependent on the applied sweep condition [18]. Increasing steam sweep ﬂow rate and pressure reduce efﬁciency signiﬁcantly, illustrating that the operational conditions of the membrane unit can not be optimized isolated from a total cycle analysis. Conditions that often improve the membrane separation, such as high pressure on the feed side and applying sweep or vacuum on the permeate side, may add to investment cost and energy consumption. Design of power-generation cycles

Figure 9.3 Oxygen-separation membrane – GT integration for production of oxygen, after [15].

9.2 Membrane Processes in Energy Systems with CO2 Capture

commonly starts from assumptions related to key units, like the gas turbine, which subsequently deﬁnes the borderlines for the other unit operations. Membrane integration with minimum cost and energy efﬁciency penalty has consequences for the operation conditions, and therefore, for the choice of membrane materials. Separating the membrane process from the fuel-gas stream necessitates few preventive actions to avoid poisoning from contaminants [19]. In the oxy-fuel process, CO2 from the combustion is recycled and could be used as the sweep gas for an integrated membrane unit to produce a stream of O2 þ CO2 for the combustor. Fuel cleanup and possible use of more stable, but less permeable membranes could add extra cost, which in this case must be considered in a total analysis. The AZEP oxy-fuel concept [20] utilizes hot exhaust gas from the natural-gas-fed combustor as sweep gas. In addition, heat is transferred across the membrane to the air stream, before the hot depleted air is expanded in the turbine to generate electricity, see Figure 9.4. The stability of the membrane limits the operation to temperatures signiﬁcantly lower than the inlet temperature of modern gas turbines (>1400 C), which reduces the cycle efﬁciency. To circumvent this limitation a combustion chamber can be introduced to heat the depleted air before entering the turbine expander. This reduces the CO2 capture rate from 100 to 85%, however, the net efﬁciency (LHV) increases from 49.6 to 53.4%, compared to 57.9% for the reference 400-MW GTCC plant without CO2 capture. The penalty of 4.5% for CO2 capture represents one of the lower values reported for thermally produced electricity, which demonstrates the gain achievable if the membrane unit could operate at higher temperature. Advantageous cost efﬁciency for oxygen membrane integration has also recently been shown by IEA for CO2 capture from 50-MW boilers [8]. Oxy-fuel

Figure 9.4 Simplified sketch of the mixed conducting oxygenseparation membrane reactor part in the AZEP concept, after [20]. Second combustor placed before the turbine improves efficiency, but also increases CO2 emission.

j201

j 9 CO

202

2

Capture with Membrane Systems

combustion employing membrane technology, and postcombustion capture with amine scrubbing, gave cost of CO2 avoidance of 22 and 70 D /tCO2, respectively. In the analysis, the capital cost for the membrane unit was estimated to 6.8 MD (assuming 1500 $/m2), anticipating an oxygen permeation rate of 1 g/m2 s (4.2 mL(STP)/cm2 min). The competitiveness of the membrane system can be explained by the lower energy penalty, and lower capital and operating costs. The potentially lower cost of oxygen membrane systems may open other medium- to small-scale applications such as the combustion of SOFC anode off-gas [21]. The efﬁciency of an integrated SOFC-GT process, for combined electrochemical and thermal electricity production, can reach more than 65% (LHV), and deliver an exhaust stream containing CO2 þ H2O only [21]. 9.2.2 Precombustion Decarbonization Processes Including Hydrogen and Carbon Dioxide Membrane Separation

Large-scale production of hydrogen for fuel and chemicals starts from fossil fuels, typically by methane steam reforming (MSR) and WGS processes. (1) Methane Steam Reforming (2) Water gas shift (3) Total reaction

CH4 þ H2O ¼ CO þ 3H2

0 DH288 ¼ 206 kJ=mol

CO þ H2O ¼ CO2 þ H2 CH4 þ 2H2O ¼ CO2 þ 4H2

0 DH288 ¼ 41 kJ=mol 0 DH288 ¼ 165 kJ=mol

The strongly endothermic equilibrium-limited steam reforming reaction is carried out at high temperature 850–900 C to reach high conversion. Water gas shift is a weakly exothermic equilibrium-limited reaction favored at low temperature. Since the yield for both reactions is limited, conversion will be governed by removing either H2 or CO2 from the reactor. Thus, steam reforming could be performed at lower and water gas shift at higher temperature, respectively, in a membrane reactor without compromising yield, see Figure 9.5. Heat is required to sustain the endothermic steam reforming reaction, which can either be supplied to the reactor externally by heaters, or internally by partial oxidation with air or oxygen. As shown in a recent study, spending some produced hydrogen as fuel for external burners, rather than natural gas, is an efﬁcient solution in routes that include CO2 capture [22]. For hydrogen-selective membranes, H2 will be obtained at lower partial pressure, and not all hydrogen can be transferred to the permeate side for subsequent use. If hydrogen is used as fuel for a gas turbine, the gas pressure has to be high, for example, 18–20 bars, which could mean that expensive compression of H2 is necessary at the permeate side. Alternatively, higher pressure applied at the feed side enables direct production of high pressure H2 at the permeate side [23]. For steam reforming, which is not favored by high pressure, the hydrogen ﬂux still increases with total pressure due to an overall increase in the partial pressure. By avoiding compression, cycle efﬁciencies above 50% (LHV) including CO2 capture are reached [24, 25]. Combustion of the remaining hydrogen and unconverted fuel in the

9.2 Membrane Processes in Energy Systems with CO2 Capture

Figure 9.5 Hydrogen-selective membrane combining steps 1 þ 2 þ 3 in precombustion decarbonization. Starting from synthesis gas, for example, after gasification of coal, step 2 þ 3 may be combined in the membrane unit. After WGS, the membrane unit may perform step 3, only.

retentate stream with pure oxygen will reduce the need for complete conversion of the fuel [26], and is a parameter to consider in a total cost and efﬁciency analysis. An elegant process to (i) provide heat to the reforming of natural gas, (ii) generate in-situ sweep gas to the WGS membrane reactor, and (iii) produce a pressurized stream of N2 þ H2O þ H2 fuel for the GT is demonstrated in the hydrogen membrane reformer (HMR) concept, see Figure 9.6 [27]. This highly integrated, high temperature membrane operation (1000–700 C), gives only a 5% reduction in cycle efﬁciency when CO2 capture and compression to 150 bar are included. For production of low-pressure H2, Middleton and coworkers have estimated the cost of CO2 avoidance in production of 230 000 Nm3/h of H2 at 1.5 bar from natural gas and coal using thin Pd membranes [28]. The total installed cost for the process gave 30% reduction relative to the baseline cost for amine scrubbing of the ﬂue gas, while the cost of CO2 avoided was 33% lower. The cost of CO2 removal, given in Figure 9.7 versus membrane permeability and speciﬁc membrane cost for precombustion routes starting from natural gas reforming, illustrates that more expensive

Figure 9.6 Three-stage membrane reactor system in the HMR concept, after [27].

j203

j 9 CO

204

2

Capture with Membrane Systems

Figure 9.7 Cost of CO2 capture for membrane reforming, after [28].

materials can be considered provided they offer increased performance. The results also demonstrated that the Pd metal cost is insigniﬁcant when the thickness reaches approximately 5 mm. The applied permeability–thickness relation versus cost of CO2 avoidance is indicative in terms of allowable cost, and also holds for other types of hydrogen-selective membranes. Extended use of hydrogen in small- and medium-size applications is expected to grow as a consequence of CO2-mitigation actions and transition to the hydrogen economy. Distributed hydrogen production via MSR þ WGS for refueling stations, employing Pd-based hydrogen-selective membranes, has recently been addressed in a techno-economic assessment [29]. The authors investigated production systems in the range 0.2–10 MW membrane reactors and found capture cost of 14 $/tCO2. Furthermore, for thin Pd alloy membranes (<10 mm), the cost of local CO2 separation (a 2 MW unit), collection in a grid of pipelines and sequestration is less than regional (40 MW) and comparable to centralized hydrogen production including CO2 sequestration, see Figure 9.8. The main cost of local production is the ﬁrst pipeline branch with low capacity, however, the cost of separation is signiﬁcantly lower than for conventional MSR. Tokyo gas has operated their Pd-based membrane reformer producing 40 Nm3/h hydrogen for more than 3000 h. The hydrogen purity obtained is 99.999%, with an energy efﬁciency of 70–76% [30]. In the case of CO2-selective membranes in precombustion processes, the fuel heating value remains at the high-pressure retentate side (H2 and unconverted fuel) in the WGS separation process. The power-cycle efﬁciency for natural-gas-fuelled GTCC including CO2-selective membranes in the WGS reactor appears less pressure dependent compared to hydrogen-selective membranes, due to the lower amount of CO2 produced in the MSR þ WGS reactions [24]. A relative simple simulation for precombustion capture made (based on CO2/H2 selectivity of 50) [24], suggests that

9.2 Membrane Processes in Energy Systems with CO2 Capture

Figure 9.8 Disaggregated cost of CO2 separation, transport and sequestration, after [29].

integration of CO2-selective membranes has approximately the same efﬁciency penalty as H2 selective membranes for typical GT inlet pressures (18–25 bar). 9.2.3 Postcombustion Capture Processes with Membrane Separation

The low pressure and CO2 concentration in postcombustion ﬂue gas streams is a demanding challenge for membrane technology. Simulations employing different membrane selectivities CO2/N2 (50–200) and ﬂue-gas compositions (10, 20, and 30% CO2) show that high CO2 recovery (>80%) is difﬁcult to achieve for a CO2 concentration below 20% [31]. Compared to conventional amine absorption capture, which requires about 4–6 GJ/tCO2, the study shows that membranes may potentially reduce this amount to 0.5 GJ/tCO2 for streams containing 20% CO2 or more [31]. The target membrane selectivity required in this case is around 60 (CO2/N2). This has already been achieved by the most promising membrane materials [32]. The concentration of CO2 in ﬂue gases originating from cement production lies between 15–30% by volume, which is higher than in ﬂue gases from power production (3–15% by volume), which could warrant CO2 removal [2]. Also in the iron and steel industry, high concentration of both CO2 and CO (20 vol.%) in the blast-furnace gas could render possible CO2 removal, and return of CO-rich gas to the furnace [2]. Based on CO2 removal for a conventional blast furnace, it is concluded that a signiﬁcant reduction of the CO2 removal costs to 17 D /tCO2 can be obtained by employing membrane technology [33]. As a consequence, CO2 removal from blast furnaces is comparable with CO2 removal in IGCC plants.

j205

j 9 CO

206

2

Capture with Membrane Systems

9.3 Properties of Membranes for Hydrogen, Oxygen, and Carbon Dioxide Separation

In this section, we will treat in more detail some critical properties related to membranes that are considered for use in CO2 capture processes. As demonstrated in the previous section, the manner of membrane integration in the different processes has major implications for the operating working conditions. For a certain membrane type, the operational window is deﬁned by the expected performance and stability. The ﬂux and selectivity are mainly assessed on basis of short-term studies in model gases. The experience with real gases and long-term studies is limited, which is reﬂected in the assessments of critical issues given below. 9.3.1 Membranes for Oxygen Separation in Precombustion Decarbonization and Oxy-Fuel Processes 9.3.1.1 Flux and Separation Ceramic membranes for oxygen separation can be divided into electrolyte type, in which only oxygen ions are mobile, and mixed-conducting types, in which both electrons and oxygen ions are mobile. The latter type can be used for pressuredriven oxygen separation, while electrolytes can be used in SOFCs for electricity production, or in the reverse mode, for oxygen pumping applying an electrical voltage. Various materials show high oxygen ion conductivity, but currently, oxides from ﬂuorite- (MO2) and perovskite- (ABO3) related families appear as being most promising [34, 35]. The similarity in materials for SOFC and oxygen membranes is also reﬂected in the common temperature of operation, being mainly between 800–1000 C [36]. Integration of these technologies as previously discussed is therefore possible in efﬁcient CO2 handling. Flux reported in the open literature rarely exceeds values of 10 mL/cm2 min, and is usually obtained at fairly low absolute oxygen pressure differential [37]. It is evident from the literature that many mixed conductors have slow surface kinetics and ﬂux in thin membranes can be limited by this effect [38]. Ambitious industrial efforts to develop oxygen-separation membranes have resulted in large progress in membrane materials, and membrane and module design. Monoliths and ﬂat structures allow high membrane packing density and can reduce the cost of production including sealing. For instance, modules with contact area of >500 m2/m3 have been produced with a checkerboard pattern of channels, and extrapolation to the AZEP process conditions expected to give an oxygen production rate of around 37 mol O2/(m3 s), or 15 MW/m3 power density [20]. According to the AZEP developers, these values correspond to targets set and conﬁrm the feasibility of the concept. Air Products (US) has adapted two ﬂat membrane concepts for synthesis gas and oxygen production. A pilot unit producing 5 tonne O2/day has been developed relying on stacked wafer-like membranes connected to a center tube collecting the oxygen [39]. Further development assumes commercial capacities (500–2000 tonne O2/day) in 2012 [17].

9.3 Properties of Membranes for Hydrogen, Oxygen, and Carbon Dioxide Separation

9.3.1.2 Stability Issues The demanding operational conditions, often combining high temperature, large gradients in total and partial pressures, and the presence of reactive-gas components challenge membrane stability. Reaction with CO2 is a problem for several good mixed-conducting perovskites (ABO3) that contain basic alkaline-earth elements on the A site. Thermodynamical stability of carbonates and oxy-carbonates is governed by lower temperatures, thus, reaction might occur during unexpected shut-downs even if avoided at the operational temperature. Acidic sulfur-containing gases (H2S, SO2) easily reacts with the same membrane components, and need therefore to be removed before the separation unit. The material designer must also consider possible evaporation of membrane components. The high temperature in combination with steam can lead to increased evaporation by metal-hydroxy components. Kinetic demixing seems to be an unavoidable phenomena originating from difference in diffusivity of the metal components in thermodynamic potential gradients [40]. The effect may lead to decomposition of the membrane, even if the membrane is thermodynamically stable in the whole oxygen partial pressure range of operation [41]. This long-term effect increases for thin membranes operating under large gradients, as encountered in synthesis gas production. Various changes in composition and morphology are often observed for membranes operated under such conditions, but this is probably due to a combination of degradation processes. The combination of high temperature and mechanical pressure also induces creep in the material, which has been studied in some oxygen-separation membranes [42]. Creep may also be a mechanism to reduce stress resulting from differences in thermal expansion of different components and chemical expansion due to reduction in oxygen content in the lattice. Chemical expansion is particularly severe in perovskites with B cations of Co and Fe. For some typical perovskites, Sirman has tabulated relative effects of various cations on A and B sites on essential membrane properties, such as oxygen ion and electronic conductivity, oxygen surface exchange rate, thermal and chemical expansion coefﬁcients, CO2 tolerance, and resistance to creep [39]. In high-ﬂux (La-alkaline earth)(Co, Fe)O3d perovskites, for example, addition of elements as Mn, Cr, Ti can improve operational stability. 9.3.2 Membranes for Hydrogen Separation in Precombustion Decarbonization

Hydrogen-separation membranes include both dense and porous types covering a temperature regime from ambient to 1000 C. We may conveniently distinguish between different types based on the thermal operational window as this is decisive for potential applications. For low-temperature polymer-based membranes, which utilize differences in solubility and diffusivity as the separation mechanism, current research is aimed at exploiting these properties [43]. In absolute terms, both glassy and rubbery polymer membranes have moderate ﬂuxes and selectivity. Although the use of crosslinked polymers has improved the performance, the complexity of implementing this approach on the large industrial scale must be solved before they ﬁnd widespread use [43]. If separation/ﬂux combination exceeding the upper

j207

j 9 CO

208

2

Capture with Membrane Systems

bound in the well-known Robeson plot is required [44], alternative membrane types are to be considered [43]. 9.3.2.1 Microporous Membranes Flux and separation Microporous inorganic membranes for gas separation mainly include microporous carbon, silica-based or related materials, and zeolite types. In the temperature regime 100–300 C, depending on material and operation condition, the presence of adsorbing components, like H2O and CO2, will hinder hydrogen diffusion leading to ﬂux reduction [45]. These membranes should preferentially work at sufﬁciently high temperature, free from surface adsorption, and with selectivity given by size exclusion. In practice, defects and a distribution in pore size result in limited selectivity dependent on molecular size. For zeolite and zeolite-like membranes, where the zeolite pore size can be controlled accurately, intercrystalline diffusion paths are difﬁcult to fully eliminate, which results in moderate separation factors [46]. High-quality microporous membranes show permeance in the range of 107–106 mol/m2 s Pa [47–50]. Amorphous silica membranes, probably the most studied and advanced microporous membrane for hydrogen separation, have a thickness in the range 20–70 nm [47, 51]. Thus, a further reduction in thickness to increase permeance, still maintaining a low defect concentration, appears as a considerable challenge [47]. A promising approach is a stage-wise sol-gel and chemical vapor deposition (CVD) synthesis process where the silica membrane obtained combines high selectivity (H2/N2 ¼ 2300) and good permeance (6.43 107 mol/m2 s Pa) [52]. Generally, high selectivity is desirable but the necessity varies with the application. For current proton exchange membrane (PEM) fuelcell applications even low CO contents in the hydrogen must be avoided due to poisoning of the anode catalyst. For combustion in, for example, gas turbines, heaters and boilers the presence of some unconverted fuel, steam and CO2 is not critical and selectivity requirements are less. For these latter applications, high ﬂux is most important, which can be increased in microporous membranes by sacriﬁcing selectivity. Microporous C, SiOC, SiON materials [49], with varying content of oxygen, have also been investigated, but currently these fall in the same ﬂux/ selectivity range as silica membranes. Hydrogen ﬂuxes in zeolite membranes are generally about 5–10 times lower than for sol-gel silica membranes due to the thicker zeolite layer needed to obtain defect-low membranes [53]. Stability issues Microporous silica membranes produced by traditional sol-gel methods are not stable in the presence of steam [51]. Different approaches have been investigated to improve the hydrothermal stability ranging from metal doping [54], inclusion of SiOC bonds in the structure to increase hydrophobicity and reduce hydroxyl formation [49], to changing to compositions mainly consisting of SiC, SiN and SiCN [55–57]. Promising results have been obtained, but the authors are not aware of results demonstrating steam stability in typical highpressure WGS or MSR conditions. On the other hand, the stability towards other WGS components (CO2/CO/CH4) and H2S is expected to be high; an advantageous

9.3 Properties of Membranes for Hydrogen, Oxygen, and Carbon Dioxide Separation

property of this membrane type. Pure-carbon membranes, however, have limited stability in some gases (CH4, H2, CO2, O2) at relevant temperatures [58], and appear less feasible for MSR and WGS processes [59]. Generally, zeolite membranes are expected to have good thermal stability, but under hydrothermal conditions the stability appears limited due to the dissolution of aluminum from the zeolite framework. Improving the hydrothermal stability seems possible, for example, by low aluminum content zeolite or titanosilicate membranes. 9.3.2.2 Dense Metal Membranes Flux and separation Dense inorganic membranes for hydrogen separation include metal, ceramic, and cermet (metal þ ceramic) types [43, 60, 61]. The metal membranes can be divided into two main groups, palladium based, and those containing Group IVB and VB metals. In addition some other metals (e.g., Ni) and amorphous phases are investigated [61]. At present, Pd-based composite membranes can be made thinner than refractory-alloy-based membranes, which in terms of ﬂux compensates for the higher permeability of the latter. For highly selective 2-mm thick Pd-23w%Ag composite membranes, a H2 ﬂux reaching 1200–1500 mL/cm2 min depending on pre-treatment at 25 bar differential pressure has been reported [62, 63], a value that corresponds to a permeance of 6.4 1031.5 102 mol/m2 s Pa0.5. The permeance is considerably reduced (5–10 times) in WGS conditions, particularly due to CO surface poisoning [62]. The refractory metals need a catalyst on the surface to enhance the kinetics of the surface reaction, and a layer of Pd or Pd-alloy is commonly applied for this purpose. The Pd layer also serves to protect the reactive refractory metal from corrosion as these easily form oxides, carbides, and nitrides. The amount of Pd coating necessary to obtain fairly stable performance [6] is marginally less than that used in state-of-the-art Pd-based composite membranes [6]. Hydrogen ﬂux reaching 423 mL/ cm2 min has been reported in H2/He feed for Pd-coated refractory metal membranes at 34 bar hydrogen differential pressure [64]. Hydrogen ﬂuxes up to 150 mL/cm2 min were achieved in WGS mixtures at pressures up to 31 bar [64]. 9.3.2.3 Stability Issues Interdiffusion between the refractory metal or porous metal support and Pd layer reduces performance and long-term stability [60, 65]. To reduce the problem, barrier layers of, for example, TiN [65], oxides [66–68] or porous PdAg [69, 70] are coated on the metal support. Investigations of Pd-based membranes in continuous sulfur-free operation have demonstrated long-term stability [30, 71–73]. Thermal cycling is more demanding due to differences in thermal and chemical (due to hydrogen dissolution) expansion between the Pd layer and support structure. The thermal expansion coefﬁcient (TEC) mismatch ((TECPd-layer TECsubstrate)/TECsubstrate) is high, >30%, for refractory metals and porous ceramic supports [64]. The expansion and contractions in refractory metals and Pd alloys induce stress that leads to deformation, wrinkles and possible detachment from the support layer [62]. Less interfacial stress is generated for thin Pd layers [74, 75], particularly on porous steel supports that have closer TEC values. Hydrogen embrittlement in metals due to hydrogen

j209

j 9 CO

210

2

Capture with Membrane Systems

dissolution is also a concern, but can be avoided by control of operation conditions and appropriate alloying. Furthermore, metal supports are also prone to creep at lower temperatures than ceramics. This could limit the total pressure differential across the membrane in MSR and WGS applications. Many fossil fuels contain sulfur components, which react with the Pd/Pd-alloy leading to ﬂux reduction by surface blocking, or even complete disintegration of the membrane. Investigations of some PdCu [76–82] and PdAu alloy [83] membranes have shown improved chemical stability towards H2S, but reports about performance in real industrial gases are meagre. Sulfur resistance appears to correlate with the Pd–Cu crystalline structure, which is determined by the operating temperature and alloy composition [79]. Failure seems to depend on H2S concentration, and not exposure time. For 125-mm thick Pd70Cu30 membranes, stable operation at 1173 K in the presence of H2S-to-H2 ratios as high as 0.0011 (1100 ppm H2S-in-H2) appears possible [84]. Under certain conditions, carbon deposition in the membrane can also occur affecting the stability [85, 86]. The many stress-generating effects, and reactive components the membrane is subjected to probably cause the commonly observed microstructural changes in thin Pd-based membranes [81, 87, 88]. Further optimization of the performance requires better understanding of these features. Alternative cermet membranes, where an interconnected Pd-based phase is conﬁned to the pores of the ceramic support may possibly offer some stability advantage, though clear evidence is lacking [89]. 9.3.2.4 Dense Ceramic Membranes Flux and separation Relatively high hydrogen permeability is found in many oxides, particularly those with soft lattices containing large basic metal ions [90, 91]. The reason is that oxygen ions move temporarily close together during vibration, allowing protons to jump from one oxygen to the next. More seldom is the combination of high mixed protonic and electronic conductivity required for pressure-driven hydrogenseparation membranes. The possibility of a non-negligible contribution of neutralhydrogen diffusion has been suggested, but further studies are needed to verify this effect [90]. Known mixed proton and electron conductor membranes require temperatures higher than 600–800 C to reach appreciable permeability. Recent publications list conductivity and some ﬂux data for several common membrane materials [90]. The maximum ﬂux reported, as far as the author know, are in the range 15–20 mL/cm2 min [92, 93]. This is in the same range as for ceramic mixed conducting oxygen-separation membranes. The addition of an electron-conducting second phase to good proton conductors, to increase the ambipolar conductivity has been reported [92, 94]. For example, by nickel addition, ﬂux through 266-mm Ba (Zr0.1Ce0.7Y0.2)O3 membranes reached nearly 1 mL/cm2 min at 900 C in pure H2. Bulk diffusion appeared rate limiting, thus lowering thickness may give an interesting ﬂux [94]. Stability issues The many stability issues discussed for ceramic oxygen-ion conductors apply also to proton conductors. Reactions with acidic gas components

9.3 Properties of Membranes for Hydrogen, Oxygen, and Carbon Dioxide Separation

and water are of similar concern for these oxides containing large often basic (Ba, Sr) elements [95]. Other issues such as kinetic demixing, creep and strength have also equal importance, but the authors are not aware of problems related to chemical expansion in this type of membranes. 9.3.3 Membranes for CO2 Separation in Precombustion Decarbonization

Recent developments demonstrate possibilities for inorganic CO2 selective membranes. Microporous membranes with strong CO2 adsorption show CO2 selectivity if other gas species are hindered in accessing the pores. For instance, at intermediate temperatures, limited CO2 selectivity to N2 (to about 400 C) and H2 (to about 200 C) is reported for MFI zeolite membranes [96]. Also, at high pressure (10–15 bars) CO2 selectivity has been demonstrated in MFI membranes (CO2/N2 separation factor 13) with promising CO2 permeance of 2.7 107 mol/m2 s Pa, though these results were obtained at 25 C [97]. A new interesting membrane type, with the potential of high-temperature operation, is the dual-phase membrane, which consists of an interconnected molten carbonate phase in a porous support [98, 99]. The electrical current loop, set up by the transport of CO2 as carbonate ions, is closed by electrical transport in the solid supporting phase. Therefore, oxygen ion conducting or metals have been used as supports to facilitate the countercurrent. It has been shown that enhanced ﬂux is obtained in the presence of oxygen on the feed side, implying that carbonate ions are the actual carrier, and not just dissolved CO2 gas. The ﬁrst few results reported show CO2/CH4 selectivity of 5, and a permeance in the order of 1 108 mol/m2 s Pa, at 500–600 C [99, 100]. 9.3.4 CO2 Separation in Postcombustion Capture 9.3.4.1 CO2 Separation Membranes CO2 capture by polymeric membranes from low-pressure ﬂue gas was early considered. The suggested necessary combination of permeability and selectivity (50 Barrer, CO2/N2 ¼ 200), however, can not be reached with existing commercial membranes [101]. Postcombustion capture has the great disadvantage compared to precombustion (20–30% CO2 at 20–50 bars) that separation is from low-pressure ﬂue gas with low CO2 concentration. If the CO2 concentration of the ﬂue gas could be increased, for instance by combustion in oxygen-enriched air, polymeric membranes may represent an alternative to amine scrubbing. Furthermore, different modiﬁcations such as mixing inorganic nanoparticles with the polymer have given enhanced membrane selectivity by increasing the solubility and the diffusivity of CO2 [102]. Dendrimer liquid membranes are also reported to have high CO2/N2 selectivities over 1000 with 1 109 m3/m2 s Pa CO2 permeance [103]. However, this immobilized liquid membrane may have insufﬁcient tolerance to handle the large pressure differences required, though recent promise has been reported [104]. These composite PAMAM dendrimer membranes are currently under evaluation, and

j211

j 9 CO

212

2

Capture with Membrane Systems

Figure 9.9 Schematic principle of the nonwetted mode of a membrane contactor.

preliminary results indicate CO2/N2 selectivity over 200 with 5 1010 m3/m2 s Pa CO2 permeance [105]. 9.3.4.2 Membrane Contactors for CO2 Capture Over the past 20 years, membrane contactors, a technology based on the combination of membrane separation and chemical absorption, have been evaluated for CO2 capture applications [106]. The nonwetting porous membrane is generally not selective, but solely acts as a barrier between the ﬂue gas and the liquid adsorbent, see Figure 9.9 [106]. Separation is determined by the reaction of one component (typically CO2 or H2S) in the gas mixture with the absorbent in the liquid. The energy-consuming regeneration of the amine solution to isolate CO2 determines to a large degree the energy required for the CO2 capture [107]. Currently, new and more energy-efﬁcient absorbents are under development, which will beneﬁt the membrane contactor technology [108]. In industry, Kvaerner Oil & Gas and W.L. Gore & Associates GmbH demonstrated the membrane contactor technology in a pilot plant at Statoils gas processing plant in Karstø on the west coast of Norway [109].

9.4 Challenges in Membrane Operation 9.4.1 Diffusion Limitation in Gas-Phase and Membrane Support

In recent decades, membrane developers have focused on developing skills to prepare thin selective membrane layers. This effort has resulted in some membranes with both high ﬂux and selectivity. Typically the ﬂux is either determined by the thickness of the selective layer, or (slow) surface kinetics. Strategies to circumvent these limitations are usually to decrease thickness and increase the surface area and/ or catalytic properties. These commonly encountered cases are typical for membranes with low to medium permeability. For highly permeable membranes, however, the gas-phase diffusion in the support or in the bulk gas may become rate limiting. In this case, the design of the membrane structure is highly important as illustrated in the two following examples.

9.4 Challenges in Membrane Operation

Figure 9.10 Hydrogen pressure drop due to depletion, concentration polarization, surface effects, transport in the palladium membrane and porous support, compared to the total hydrogen partial pressure drop. (a) H2 : N2 ¼ 50 : 50; (b) H2 : N2 : CO2 ¼ 50 : 25 : 25;

(c) H2:N2:CO ¼ 50 : 25 : 5; (d) H2 : CO2 : H2O : CO : CH4 ¼ 60 : 19 : 16 : 4 : 1. Pfeed ¼ 20 bar, T ¼ 400 C. Depletion means the lowering of H2 bulk gas concentration due to H2 removal along the tube length. Data after [62].

(i) Highly permeable 1–3-mm thick Pd-23 wt% hydrogen-selective membranes supported on 0.48-mm thick porous stainless steel tubes with 2-mm pore size have been reported with a pure H2 permeance of 6.4 103 mol/m2 s Pa0.5 [62]. Operation of these membranes in gas mixtures, for example, H2 þ N2 (N2 assumed to behave as an inert) suggests that the hydrogen ﬂux is mainly limited by a gas-phase diffusion limitation at the feed side. A hydrogen-depleted concentration-polarization layer is built up, reducing the efﬁcient partial pressure of hydrogen, and thereby also the gradient in pressure sustaining the ﬂux. Figure 9.10 illustrates the estimated partial pressure drop for sustaining the ﬂux by three major processes; gas diffusion to the membrane feed surface, transport through the PdAg 23 wt.%, and transport through the porous steel support, respectively. The gas-phase limitations imply a need for improving membrane and module design, and an optimization of feed-ﬂow conditions to reduce the thickness of the hydrogen-depleted layer. The example using a Pdbased membrane may also be used to illustrate the problem of surface reaction rate limitation. Adsorption of other gas molecules on the surface hinders H2 incorporation, and therefore reduces ﬂux. The effect is particularly strong for CO, which is illustrated in Figure 9.10. A comparison with the high-ﬂux situation with only an inert molecule present shows that the importance of the gas-phase diffusion limitation is drastically reduced. This illustrates that the operation of highly permeable Pd-based membranes (or Group IVand V membranes with Pdcatalyst layer) in WGS conditions is strongly limited by a combination of surface effects and gas-phase diffusion limitations. Thus, if the surface effect could be

j213

j 9 CO

214

2

Capture with Membrane Systems

Figure 9.11 2D pressure profiles (isobars) and 2D flux profiles (arrows) as a function of permeance (F) of microporous silica membranes on the inside wall of multichannel supports. Very low F (a), value for the state-of-the-art (b), and very high F (c), after [110].

reduced, the expected ﬂux increase will be limited by gas-phase diffusion. These design and operational implications set by gas-phase diffusion limitations are not limited to the case of Pd-based membranes, but to all highly permeable gas-separation membranes. (ii) For highly permeable separation layers the resistance of the support structure must be considered. Hollow ﬁbers, multichannel and honeycomb elements all have high surface area per volume, and as such represent possible membrane designs for cheap large-scale gas-separation systems. While hollow ﬁbers and honeycomb structures divide the feed and permeate streams by similar separating membrane walls, the distance from the feed to the permeate side in common multichannel elements varies considerable depending on channel position. The ﬂux per area efﬁciency of the multichannel element thus depends on the net contribution from all channels, that is, the resistance of the support must be insigniﬁcant compared to the resistance of the selective layer [110]. This is illustrated in Figure 9.11 where the contributions from channel 3 (inner channel) and 2 diminish as the permeability of the membrane layer on the inside channel wall increases. The permeance of highly permeable membranes is typically found in the region 106 mol/m2 s Pa, where the resistance from commercially available multichannel supports inﬂuence the efﬁciency. 9.4.2 Membrane Module Design and Catalyst Integration

The mechanical properties of the membrane are essential in operation and module design. For instance, hollow carbon ﬁbers fabricated by pyrolysis of polymers are seemingly too brittle for practical applications [111]. Ceramic capillaries prepared by extrusion are much stronger, but appear limited in maximum length due to

9.4 Challenges in Membrane Operation

vibrations that might occur during operation. To improve the resistance to mechanical stress one faces the dilemma of increasing the wall thickness and/or reducing porosity and/or pore size; all strategies that will increase support resistance. The optimal capillary design is therefore a compromise between sufﬁciently mechanical strength and permeance. Monolithic membranes with thin walls offer probably the most stable and efﬁcient design, the latter not only due to the high surface area, but also because of the small comparative sealing area. Several designs have been suggested and recently summarized by Carolan [112]. It should, however, be noted that monoliths that provides possibility of crossﬂow of two separate gas streams give complex manifold systems that may not be easy to fabricate [20]. Ideally, the channels should be made sufﬁciently small to reduce the gas-phase diffusion limitation, but at the same time not generate too high a crossﬂow pressure drop. The ﬂat design, which in different forms has been widely investigated in planar SOFC stacks, also provides a means to create high membrane surface area and narrow gas-ﬂow paths. Stacks of ﬂat membranes using spacers to control the distance between membrane plates have been developed by Air Products and partners [113] for their ceramic oxygen-separation membranes in synthesis gas production. The wafer-like design operates with the membrane in mechanical compression between an outer porous support and an internal core of microchannels that distributes the gas evenly. The low-pressure stream is conﬁned to the internal channels, while the pressurized gas is streaming between the wafers. The planar designs reported for synthesis gas and oxygen production [39, 112] limit the extent of necessary metal–ceramic seals for integration in steel housings, but require high-temperature ceramic–ceramic seals to connect the wafers in the stack. Extensive ceramic–ceramic sealing is also demonstrated for hollow ﬁbers, which can be bundled together and sealed to ceramic end sealings [51]. The two main reactions, discussed in this chapter, MSR and WGS, require catalyst and operational control of mass and heat ﬂow. For the highly endothermic MSR reaction, heat is traditionally provided by natural-gas burners, which would require additional systems for CO2 capture. More elegant is in-situ oxidation, for example, as done in the previously discussed hydrogen membrane reformer (HMR) [27]. Alternatively, by employing an oxygen membrane providing oxygen to the fuel side, partial oxidation (exothermic) and steam reforming can be combined to control temperature [14]. The reactor design for these processes should aim at reducing mechanical stress originating from thermal and chemical expansion mismatch, total pressure and temperature gradients. Close integration of the catalyst in MSR to ensure sufﬁcient heat transfer, could be a complicating factor if the chemical compatibility of catalyst and membranes is not sufﬁcient, or if the membrane and catalyst have different lifetimes. For monolithic structures the compatibility issue appears more important than for tubes or plates where the catalyst can be placed more easily externally to the membrane surface, while still being in close proximity. The WGS reaction is only weakly exothermic providing higher ﬂexibility in catalyst integration. Middleton and coworkers suggested that in the WGS process, the reactor containing the catalyst could be separated from the membrane separation unit, in a three-stage sequential process of reaction and separation [114]. There are several advantageous with this concept, (i) each reactor and separation step can be optimized

j215

j 9 CO

216

2

Capture with Membrane Systems

with respect to sizing, design, and to a certain degree, temperature, (ii) feed ﬂow rate and sweep gas can be optimized in each separation step, (iii) the exchange of membranes and catalyst can be done independently, (iv) problems related to chemical noncompatibility of catalyst and membrane can be eliminated. The downside of the concept is an increase in catalyst volume (33%) and membrane area (29%) compared to a single-stage catalytic membrane reactor process designed for capture of 2 Mtonne/yr CO2 in the production of hydrogen by authothermal reforming and WGS at Grangemouth reﬁnery in Scotland [114].

9.5 Concluding Remarks

Increasing awareness about environmentally related problems has led to large efforts for developing clean and energy-efﬁcient technology. In this chapter we have given examples demonstrating the many opportunities, offered by emerging membrane technology, to efﬁciently solve key problems related to GHG emission control. The encouraging involvement of industry and public funding organization ensures faster realization, as well as illustrating the competitiveness of the technology. Sufﬁcient ﬂux and selectivity capacity is reported for several existing membrane systems, though long-term performance veriﬁcation is less clear. Further R&D efforts are therefore still needed to further verify these critical parameters, and particularly, considerable more attention should be directed to stability issues. Optimization of the membrane operation is though a compromise that includes design of the membrane and module on the one hand, and process integration deciding operation condition on the other. Introduction of O2-, H2-, and CO2-selective membranes in large-scale CO2-capture processes is still some years into the future. In the coming years membrane development will continue on a broad basis, and novel ways of process integration will evolve that strengthens the future impact of membrane technology in CO2 mitigation.

Acknowledgments

The support from the Research Council of Norway through the NANOMAT projects Functional Oxides for Energy Technology (Grant No. 15 851/431) and Materials for Hydrogen Technology (Grant No. 158 516/S10) is gratefully acknowledged.

References 1 Davison, J. (2007) Energy, 32, 1163. 2 IPCC (2005) IPCC Special Report onCarbon Dioxide Capture and Storage, Prepared by

Working Group III of the Intergovernmental Panel on Climate Change (eds B. Metz, O. Davidson, H.C. de Coninck, M. Loos and

References

3

4

5

6

7

8

9

10

11

12 13

14

L.A. Meyer), Cambridge University Press, United Kingdom and New York, NY, USA, p. 442. IAE – International Energy Agency (2006) Key World Energy Statistics 2006, IAE - International Energy Aogency. Steeneveldt, R., Berger, B. and Torp, T.A. (2006) Chemical Engineering Research & Design, 84, 739. (2007) European Technology Platform for Zero Emission Fossil Fuel Power Plants http://www.zero-emissionplatform.eu/ website/docs/ETP%20ZEP/EU% 20Flagship%20Programme%20for% 20CCS.pdf (2007) Alstom Environment Control Systems Alstom and Statoil to jointly develop project for chilled ammonia-based CO2 capture for natural gas in Norway. International Energy Agency (IEA) (2007) Near-term opportunities for carbon dioxide capture and storage. Global Assessments Workshop. IEA Greenhouse Gas R&D Programme (IEA GHG) (2007) CO2 capture from medium scale combustion installations 2007/7. IEA Greenhouse Gas R&D Programme (IEA GHG) (2007) Distributed collection of CO2, 2007/12, September 2007. Kather, A., Mieske, K., Hermsdorf, C., Klostermann, M. and K€opke, D. (2006) 8th International Conference on Greenhouse Gas Control Technologies (GHGT-8), Trondheim, Norway. Allam, R.J., Bredesen, R. and Drioli, E. (2003) in Carbon Dioxide Recovery and Utilization (ed. M. Aresta), Kluwer, Dordrecht, pp. 55–118. Rubin, E.S., Chen, C. and Rao, A.B. (2007) Energy Policy, 35, 4444. Rubin, E.S., Yeh, S., Antes, M., Berkenpas, M. and Davison, J. (2007) International Journal of Greenhouse Gas Control, 1, 188. Bredesen, R. and Sogge, J. (1996) Presentation at the United Nation Seminar on the Ecological Applications of Innovative Membrane Technology in the Chemical Industry, Cetraro, Calabria, Italy.

15 Stiegel, G.J. (1999) Membrane Technology, 110, 5. 16 Vente, J.F., Haije, W.G. and Rak, Z.S. (2006) Journal of Membrane Science, 276, 178. 17 Armstrong, P.A., Bennet, D.L., Foster, E.P.T. and Stein, V.E. (2005) Presentation at Gasiﬁcation Technologies 2005, San Francisco, USA. 18 Sander, F., Foeste, S. and Span, R. (2006) (eds R. Bredesen and H. Ræder), Proceedings of the 9th International Conference on Inorganic Membranes, Lillehammer, Norway, pp. 224–228. 19 Trembly, J.P., Gemmen, R.S. and Bayless, D.J. (2007) Journal of Power Sources, 163, 986. 20 Sundkvist, S.G., Julsrud, S., Vigeland, B., Naas, T., Budd, M., Leistner, H. and Winkler, D. (2007) International Journal of Greenhouse Gas Control, 1, 180. 21 Maurstad, O., Bredesen, R., Bolland, A., Kvamsdal, H.M. and Schell, M. (2004) in 7th International Conference on Greenhouse Gas Control Technologies (GHGT-7), Vancouver, Canada (eds E.S. Rubin, E.W. Keith and C.F. Gilboy), Elsevier Science, Oxford, UK. 22 Feng, W., Ji, P.J. and Tan, T.W. (2007) AIChE Journal, 53, 249. 23 Jordal, K., Bredesen, R., Kvamsdal, H.M. and Bolland, O. (2004) Energy, 29, 1269. 24 Kaggerud, K., Gjerset, M., Mejdell, T., Kumakiri, I., Bolland, A. and Bredesen, R. (2005) Greenhouse Gas Control Technologies, 7, 1857. 25 Kvamsdal, H.M., Jordal, K. and Bolland, O. (2007) Energy, 32, 10. 26 Johannessen, E. and Jordal, K. (2005) Energy Conversion and Management, 46, 1059. 27 Åsen, K., Vigeland, B., Norby, T., Larring, Y. and Mejdell, T. (2004) in 7th International Conference on Greenhouse Gas Control Technologies (GHGT-7), Vancouver, Canada (eds E.S. Rubin, E.W. Keith and C.F. Gilboy), Elsevier Science, Oxford, UK. 28 Middleton, P., Solgaard-Andersen, H. and Rostrup-Nielsen, H.T. (2002) World

j217

j 9 CO

218

29 30 31

32 33

34

35 36

37

38

39

40

41 42 43 44

2

Capture with Membrane Systems

Hydrogen Energy Conference, Montreal, Canada. Sjardin, M., Damen, K.J. and Faaij, A.P.C. (2006) Energy, 31, 2523. Yasuda, I. and Shirasaki, Y. (2007) Materials Science Forum, 539–543, 1403. Bounaceur, R., Lape, N., Roizard, D., Vallieres, C. and Favre, E. (2006) Energy, 31, 2556. Favre, E. (2007) Journal of Membrane Science, 294, 50. Lie, J.A., Vassbotn, T., Hagg, M.B., Grainger, D., Kim, T.J. and Mejdell, T. (2007) International Journal of Greenhouse Gas Control, 1, 309. Steele, B.C.H. (1989) in High Conductivity Solid Ionic Conductors, Recent Trends and Applications (ed. T. Takahashi), World Scientiﬁc, Singapore, pp. 402–446. Kilner, J.A. (2000) Solid State Ionics, 129, 13. Fontaine, M.-L., Norby, T., Larring, Y., Grande, T. and Bredesen, R. (2008) in Inorganic Membranes: Synthesis, Characterization and Applications (eds R. Mallada and M. Menendez), Elsevier, Amsterdam. Vente, J.F., McIntosh, S., Haije, W.G. and Bouwmeester, H.J.M. (2006) Journal of Solid State Electrochemistry, 10, 581. Bouwmeester, H.J.M., Kruidhof, H. and Burggraaf, A.J. (1994) Solid State Ionics, 72, 185. Sirman, J. (2006) in Nonporous Inorganic Membranes for Chemical Processing (eds F. Sander and M.V. Mundschau), WileyVCH, Weinheim, pp. 165–184. Stølen, S. and Grande, T. (2004) in Chemical Thermodynamics of Materials (eds S. Stølen and T. Grande), John Wiley & Sons, Ltd, New York, pp. 1–27. Schmalzried, H. and Laqua, W. (1981) Oxidation of Metals, 15, 339. Majkic, G., Wheeler, L. and Salama, K. (2000) Acta Materialia, 48, 1907. Ockwig, N.W. and Nenoff, T.M. (2007) Chemical Reviews, 107, 4078. Robeson, L.M. (1991) Journal of Membrane Science, 62, 165.

45 Hong, M., Li, S., Falconer, J.L. and Noble, R.D. (2008) Journal of Membrane Science, 298, 182. 46 McLeary, E.E., Jansen, J.C. and Kapteijn, F. (2006) Microporous and Mesoporous Materials, 90, 198. 47 Verweij, H., Lin, Y.S. and Dong, J.H. (2006) MRS Bulletin, 31, 756. 48 Mottern, M.L., Shqau, K., Shi, J.Y., Yu, D. and Verweij, H. (2007) International Journal of Hydrogen Energy, 32, 3713. 49 de Vos, R.M., Maier, W.F. and Verweij, H. (1999) Journal of Membrane Science, 158, 277. 50 Peters, T.A., Fontalvo, J., Vorstman, M.A.G., Benes, N.E., van Dam, R.A., Vroon, Z.A.E.P., van Soest-Vercammen, E.L.J. and Keurentjes, J.T.F. (2005) Journal of Membrane Science, 248, 73. 51 Yoshino, Y., Suzuki, T., Nair, B.N., Taguchi, H. and Itoh, N. (2005) Journal of Membrane Science, 267, 8. 52 Gopalakrishnan, S., Yoshino, Y., Nomura, M., Nair, B.N. and Nakao, S.I. (2007) Journal of Membrane Science, 297, 5. 53 Gu, X., Tang, Z. and Dong, J. (2008) Microporous and Mesoporous Materials, 111, 441. 54 Kanezashi, M. and Asaeda, M. (2006) Journal of Membrane Science, 271, 86. 55 Ciora, R.J., Fayyaz, B., Liu, P.K.T., Suwanmethanond, V., Mallada, R., Sahimi, M. and Tsotsis, T.T. (2004) Chemical Engineering Science, 59, 4957. 56 Iwamoto, Y., Sato, K., Kato, T., Inada, T. and Kubo, Y. (2005) Journal of the European Ceramic Society, 25, 257. 57 Volger, K.W., Hauser, R., Kroke, E., Riedel, R., Ikuhara, Y.H. and Iwamoto, Y. (2006) Journal of the Ceramic Society of Japan, 114, 567. 58 Koresh, J.E. and Soffer, A. (1987) Separation Science and Technology, 22, 973. 59 Lu, G.Q., Diniz da Costa, J.C., Duke, M., Giessler, S., Socolow, R., Williams, R.H. and Kreutz, T. (2007) Journal of Colloid and Interface Science, 314, 589. 60 Dolan, M.D., Dave, N.C., Ilyushechkin, A.Y., Morpeth, L.D. and McLennan, K.G.

References

61

62

63

64

65 66

67 68

69

70

71

72

73

(2006) Journal of Membrane Science, 285, 30. Phair, J.W. and Donelson, R. (2006) Industrial & Engineering Chemistry Research, 45, 5657. Peters, T.A., Stange, M., Klette, H. and Bredesen, R. (2008) Journal of Membrane Science, 316, 119. Peters, T.A., Stange, M., Bredesen, R. (2008) Thin Pd-23w%Ag membranes for hydrogen separation, Presentation at the 10th International conference on inorganic membranes. (ICIM-10, August 18–22, 2008), Tokyo, Japan. Mundschau, M.V., Xie, X. and Evenson, C.R. (2006) (eds F. Sander and M.V. Mundschau), Nonporous Inorganic Membranes for Chemical Processing. Wiley-VCH, Weinheim, pp. 107–138. Nam, S.E. and Lee, K.H. (2005) Industrial & Engineering Chemistry Research, 44, 100. Yepes, D., Cornaglia, L.M., Irusta, S. and Lombardo, E.A. (2006) Journal of Membrane Science, 274, 92. Huang, Y. and Dittmeyer, R. (2006) Journal of Membrane Science, 282, 296. Su, C.L., Jin, T., Kuraoka, K., Matsumura, Y. and Yazawa, T. (2005) Industrial & Engineering Chemistry Research, 44, 3053. Ma, Y.H., Akis, B.C., Ayturk, M.E., Guazzone, F., Engwall, E.E. and Mardilovich, I.P. (2004) Industrial & Engineering Chemistry Research, 43, 2936. Ayturk, A.E., Mardilovich, I.P., Engwall, E.E. and Ma, Y.H. (2006) Journal of Membrane Science, 285, 385. Tosti, S., Basile, A., Bettinali, L., Borgognoni, F., Chiaravalloti, F. and Gallucci, F. (2006) Journal of Membrane Science, 284, 393. Matzakos, A. (2006) Presentation at the 9th International Conference on Inorganic Membranes, Lillehammer, Norway. Pex, P.P.A.C., Delft, Y.C.v., Correia, L.A., Veen, H.M.v., Jansen, D. and Dijkstra, J.W. (2004) in 7th International Conference on Greenhouse Gas Control Technologies (GHGT-7), Vancouver, Canada (eds E.S.

74

75

76 77 78

79

80

81 82

83

84

85

86

87

Rubin, E.W. Keith and C.F. Gilboy), Elsevier Science, Oxford, UK. Tosti, S., Bettinali, L., Castelli, S., Sarto, F., Scaglione, S. and Violante, V. (2002) Journal of Membrane Science, 196, 241. Guazzone, F., Payzant, E.A., Speakman, S.A. and Ma, Y.H. (2006) Industrial & Engineering Chemistry Research, 45, 8145. Roa, F., Block, M.J. and Way, J.D. (2002) Desalination, 147, 411. Alfonso, D.R., Cugini, A.V. and Sholl, D.S. (2003) Surface Science, 546, 12. Kulprathipanja, A., Alptekin, G.O., Falconer, J.L. and Way, J.D. (2004) Industrial & Engineering Chemistry Research, 43, 4188. Kulprathipanja, A., Alptekin, G.O., Falconer, J.L. and Way, J.D. (2005) Journal of Membrane Science, 254, 49. Grimmer, P.J., Xie, X., Evenson, C.R., Mundschau, M.V. and Wright, H.A. (2006) Presentation at the 23th Annual International Pittsburgh Coal Conference, Pittsburgh, USA. Yuan, L., Goldbach, A. and Xu, H. (2007) Journal of Physical Chemistry. B, 111, 10952. Morreale, B.D., Ciocco, M.V., Howard, B.H., Killmeyer, R.P., Cugini, A.V. and Enick, R.M. (2004) Journal of Membrane Science, 241, 219. Gade, S.K., Keeling, M.K., Steele, D.K., Thoen, P.M., Way, J.D., DeVoss, S. and Alptekin, G.O. (2006) in Proceedings of the 9th International Conference on Inorganic Membranes, Lillehammer, Norway (eds R. Bredesen and H. Ræder), pp. 112–117. Iyoha, O., Enick, R., Killmeyer, R. and Morreale, B. (2007) Journal of Membrane Science, 305, 77. Li, H., Goldbach, A., Li, W. and Xu, H. (2007) Journal of Membrane Science, 299, 130. Gao, H.Y., Lin, Y.S., Li, Y.D. and Zhang, B.Q. (2004) Industrial & Engineering Chemistry Research, 43, 6920. Shu, J., Bongondo, B.E.W., Grandjean, B.P.A. and Kaliaguine, S. (1997) Journal of Materials Science Letters, 16, 294.

j219

j 9 CO

220

2

Capture with Membrane Systems

88 Mekonnen, W., Arstad, B., Klette, H., Walmsley, J.C., Bredesen, R., Venvik, H. and Holmestad, R., Journal of Membrane Science, 310, 337. 89 Balachandran, U., Lee, T.H., Chen, L., Song, S.J., Picciolo, J.J. and Dorris, S.E. (2006) Fuel, 85, 150. 90 Norby, T. and Haugsrud, R. (2006) in Nonporous Inorganic Membranes for Chemical Processing (eds F. Sander and M.V. Mundschau), Wiley-VCH, Weinheim, pp. 1–48. 91 Fontaine, M.L., Larring, Y., Norby, T., Grande, T. and Bredesen, R. (2007) Annales de Chimie-Science des Materiaux, 32, 197. 92 Hamakawa, S., Li, L., Li, A. and Iglesia, E. (2002) Solid State Ionics, 148, 71. 93 Vigeland, B. and Åsen, K. (2006) 8th International Conference on Greenhouse Gas Control Technologies (GHGT-8), Trondheim, Norway. 94 Zuo, C.D., Lee, T.H., Dorris, S.E., Balachandran, U. and Liu, M.L. (2006) Journal of Power Sources, 159, 1291. 95 Kreuer, K.D. (1997) Solid State Ionics, 97, 1. 96 Kumakiri, I., Lecerf, N. and Bredesen, R. (2004) Transactions of the Materials Research Society of Japan, 29, 3271. 97 Sebastian, V., Kumakiri, I., Bredesen, R. and Menendez, M. (2007) Journal of Membrane Science, 292, 92. 98 Chung, S.J., Park, J.H., Li, D., Ida, J.I., Kumakiri, I. and Lin, J.Y.S. (2005) Industrial & Engineering Chemistry Research, 44, 7999. 99 Kawamura, H., Yamaguchi, T., Nair, B.N., Nakagawa, K. and Nakao, S. (2005) Journal of Chemical Engineering of Japan, 38, 322. 100 Yamaguchi, T., Niitsuma, T., Nair, B.N. and Nakagawa, K. (2007) Journal of Membrane Science, 294, 16. 101 Powell, C.E. and Qiao, G.G. (2006) Journal of Membrane Science, 279, 1. 102 Zimmerman, C.M., Singh, A. and Koros, W.J. (1997) Journal of Membrane Science, 137, 145.

103 Kovvali, A.S. and Sirkar, K.K. (2001) Industrial & Engineering Chemistry Research, 40, 2502. 104 Duan, S.H., Kouketsu, T., Kazama, S. and Yamada, K. (2006) Journal of Membrane Science, 283, 2. 105 Kazama, S., Kai, T., Taniguchi, I., Duan, S., Chowdhury, F. and Fujioka, Y. (2007) Presentation at 4th Trondheim Conference on CO2 Capture, Transport and Storage. 106 Feron, P.H.M., Jansen, A.E. and Klaassen, R. (1992) Energy Conversion and Management, 33, 421. 107 Abu-Zahra, M.R.M., Niederer, J.P.M., Feron, P.H.M. and Versteeg, G. (2006) 8th International Conference on Greenhouse Gas Control Technologies (GHGT-8), Trondheim, Norway. 108 Mamun, S., Svendsen, H.F., Hoff, K.A. and Juliussen, O. (2007) Energy Conversion and Management, 48, 251. 109 International Energy Agency (IEA) (2003) Zero Emission Technologies for Fossil Fuels Technology Status Report. Working Party on Fossil Fuels. 110 Zivkovic, T., Benes, N.E. and Bouwmeester, H.J.M. (2004) Journal of Membrane Science, 236, 101. 111 Liang, C., Sha, G. and Guo, S. (1999) Carbon, 37, 1391. 112 Carolan, M. (2006) in Nonporous Inorganic Membranes for Chemical Processing (eds F. Sander and M.V. Mundschau), WileyVCH, Weinheim, pp. 215–244. 113 Allam, R., White, V., Stein, V., McDonald, C., Ivens, N. and Simmonds, M. (2005) Carbon Dioxide Capture for Storage in Deep Geologic Formations, Elsevier Science, Amsterdam, pp. 513–535. 114 Middleton, P., Klette, H., Bredesen, R., Larring, Y., Ræder, H. and Lowe, C. (2004) in 7th International Conference on Greenhouse Gas Control Technologies (GHGT-7), Vancouver, Canada (eds E.S. Rubin, E.W. Keith and C.F. Gilboy), Elsevier Science, Oxford, UK.

j221

10 Seawater and Brackish-Water Desalination with Membrane Operations Raphael Semiat and David Hasson

10.1 Introduction: The Need for Water

According to UN reports, between 20–25% of the worlds population do not have access to good-quality water. People are dying daily due to illnesses related to poorquality water. The availability of drinking water is continuously decreasing due to the over-usage of aquifers and traditional water sources. This is due in part to how humankind treats the environment, resulting in the pollution of water resources. This is causing people to concentrate in large cities where they expect a better life. These cities are also starting to suffer from a lack of natural, good-quality water. In many places, people are responsible for getting their own water from a distance, wasting considerable time and effort in fulﬁlling this important task. In many places farmers are dying of hunger since they lack both the technique and the capability to pump water from a nearby river to irrigate their crops. Over 98% of water sources on earth are undrinkable due to salt content. Only a fraction of the good-quality water is actually used due to the naturally uneven distribution of the water. The problem of water shortage is not only a problem of proper techniques; it is also a social and educational problem depending in many cases on national and international efforts as well as on technical solutions. We need better techniques to provide good-quality water at a low cost, and we must educate people to make better usage of this cheap, yet very costly, resource. The aim of this chapter is to deal with some of the best available water production and puriﬁcation techniques, as well as discuss desalination issues based on membranes. Increasing production at affordable costs is one of humankinds most important objectives.

10.2 Membrane Techniques in Water Treatment

The term desalination has lately started to include diverse treatments to purify different water sources, from slightly polluted water, through wastewater and Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

222

brackish water, up to seawater. Membrane techniques are used in many ways to improve water quality. reverse osmosis (RO) is currently the fastest growing desalination technique in industry, emerging even faster than evaporation techniques. Electrodialysis is used for the treatment of slightly polluted water. Other types of membranes are used in different techniques to remove suspended and dissolved matter from raw waters. The main pretreatment steps before using RO membranes are based on the removal of suspended matter from feed water, sometimes including disinfection substances to kill bacteria, followed by a means to remove organic matter and chlorine compounds by active carbon, acidulation to remove carbonate, and more. The feed water is then pumped to an elevated pressure, high enough to overcome the osmotic pressure of the salt-concentrated solution resulting from the actual product recovery of the feed water. Other techniques based on water evaporation are also used for desalination, yet are not included in the scope of this chapter [1]. Osmotic pressure is a property of a solution containing dissolved matter, such as salts, starch or sugar in water; the latter are similar to materials existing in the roots of most plants. The relatively high concentration enables transferring water from the soil surrounding the root through a membrane at the skin of the root. Applying increased pressure to such a concentrated solution behind the membrane reduces water passage and may stop the ﬂow of water (the pressure level that stops the ﬂow is deﬁned as the osmotic pressure of the solution). Higher pressure applied on the solution side of a synthetic membrane, well above the osmotic pressure, will overpower the solutions properties and transfer water from the concentrated solution through the membrane in a direction opposite to the natural action at the plant root. This is the basis of the reverse-osmosis process: it enables selective water permeation through a membrane from the saline side to the freshwater side [2]. Salts rejected by the membrane stay in the concentrating stream but are continuously disposed from the membrane module by fresh feed to maintain the separation. Continuous removal of the permeate product enables the production of freshwater. RO membrane-building materials are usually polymers, such as cellulose acetates, polyamides or polyimides. The membranes are semipermeable, made of thin 30–200 nanometer thick layers adhering to a thicker porous support layer. Several types exist, such as symmetric, asymmetric, and thin-ﬁlm composite membranes, depending on the membrane structure. They are usually built as envelopes made of pairs of long sheets separated by spacers, and are spirally wound around the product tube. In some cases, tubular, capillary, and even hollow-ﬁber membranes are used. Water passage through reverse-osmosis membranes is based on water dissolution in the membrane walls followed by diffusion to the other side of the membrane. RO membranes are denser membranes, containing almost no holes. The membrane skin, supported by a porous polymeric layer, is responsible for the membrane properties. The solubility of water in the membrane is much higher than the solubility of the salts present in feed water, hence enabling the separation between water and salt ions, which are also relatively large molecules surrounded by water molecules. The integrity of the skin layer is very important for the rejection of salts. Scratches and holes in the skin enhance the passage of salt ions and thus reduce salt rejection. A SEM picture of a RO membrane is presented in Figure 10.1. The skin,

10.2 Membrane Techniques in Water Treatment

Figure 10.1 The structure of a RO membrane – the thin skin is responsible for membrane flux and rejection properties. Picture taken from Blanco et al. [53].

presented at the top of the picture, is responsible for the membrane properties. Membranes are spirally wounded as shown in Figure 10.2 (800 and 1600 in diameter) and are inserted into pressure vessels. Ultraﬁltration and microﬁltration membranes produce high porosities and pore sizes in the range of 30–100 nanometers (UF) and higher (MF), which enable the passage of larger dissolved particles and even some suspended particles. The separation-ﬁltration mechanism is based on molecule/particle sizes. The nanoﬁltration membrane lies between the UF and RO membranes, combining the properties of both so that the two mechanisms coexist. In addition, the NF membrane may be

Figure 10.2 Spiral-wound RO membranes, 800 and 1600 in diameter. Cooperation between Nitto Denco/Hydranautics and Graham Tec.

j223

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

224

charged electrically, depending on functional groups acting on the membrane surfaces. This charge affects the passage of molecules through the membranes. The membranes may be found in different types of modules, enabling their use in water puriﬁcation and treatment. Wilbert et al. [3] described various treatments available for surface water and other sources. Nanoﬁltration membranes are used to remove hardness from drinking water [4, 5]. They may also be used to remove other unwanted dissolved species, even the partial removal of nitrates from ground water. It was recently shown that RO and NF membranes may be backwashed by direct osmotic pressure to clean membrane surfaces, a simple and very beneﬁcial technique [6, 7]. Ultraﬁltration and microﬁltration can be backwashed occasionally to remove accumulated solids from membranes. UF and MF membranes may be used to remove micrometer-sized and upper suspended particles, namely bacteria, algae, and so on, they can also be used to remove Guardia and Cryptosporidium, as well as most viruses found in surface water. In fact, the solid layer (cake) adhering to the membranes in the latter two techniques acts like a dynamic membrane [8, 9], removing smaller particles even at colloidal and virus levels. The use of MF membranes may be cheaper than sand ﬁltration in the treatment of surface water. The international water company, Ondeo (Lyonnaise des Eaux), uses MF membranes combined with active coal and sedimentation stages to purify polluted Seine River water for drinking purposes [10]. Veolia also uses MF combined with NF to get good-quality water. Many other companies, membrane manufacturers or users are involved in producing clean wastewater, either directly together with a MBR bioreactor [11] or using membranes after they have passed through the bioreactor. In Singapore, wastewater is treated with UF and RO membranes to make NewWater for usage in microelectronic fabrication [12]. Part of the water is mixed with surface water for regular usage. More on wastewater treatment is provided below. Electrodialysis (ED), or reversible electrodialysis (RED), involves applying a DC electrical ﬁeld across a membranes stack. Ions are transferred through semipermeable membranes into concentrated streams, leaving behind a diluted salt solution. This was considered a promising technique mainly because of the relative insensitivity of the membranes for fouling, and due to the thermodynamic transfer properties of this technique. Unfortunately, the technique did not succeed in taking its naturally expected position among other processes. It is currently used primarily for brackish-water desalination and water puriﬁcation [13]. EDR membranes are also used to remove special salts, such as nitrates, from slightly polluted water. Strathmann [14] provides a cost estimate of the ED process. The use of membranes is inﬁltrating into the process industry, where improved water quality is needed. Power stations, petrochemical and high-tech production plants are seeking improved water quality and are using different types of membranes to meet their needs. Additional information on different aspects of desalination processes was reported by Semiat [1]. Electrical power is the energy source for RO desalination. A reverse-osmosis desalination plant is presented schematically in Figure 10.3. Electricity is used to

10.2 Membrane Techniques in Water Treatment

Figure 10.3 Schematic presentation of a reverse-osmosis desalination plant.

pump the water at a relatively high operating pressure. The product penetrates the membrane and exits at a predesigned recovery, deﬁned as the product-to-feed ratio. The high-pressure purged concentrate contains energy that may be recovered using turbines or pressure-exchange devices [15]. The osmotic pressure of seawater, for example, varies from 24 bars to twice as much for concentrate at 50% recovery. Operating pressures therefore vary between 10–25 bars for brackish water and 60–80 bars for seawater in order to allow sufﬁcient permeation at relatively high concentrations of the brine along the pressure vessel. The process takes place at ambient temperatures. Water conversion can increase to 70–95% recovery in the case of contaminated or brackish water, or 35–50% recovery using seawater. The low recovery from seawater is due to the high osmotic pressure of the concentrate leaving the membrane modules, depending on the recovery ratio and the need to operate at higher pressure, where investment increases signiﬁcantly with operating pressure. Lower water recovery is obtained in relatively closed water bodies, such as the Red Sea or the Persian Gulf, due to higher salt concentrations. Water temperature inﬂuences membrane performance. Flux through a membrane increases with water temperature and is bounded by membrane limitations, yet salt rejection and product quality are reduced with an increase in water temperature. Hot seawater ﬂowing from the cooling system of a large power plant may increase efﬁciency at the expense of water quality. The quality of the water produced depends on membrane-rejection properties together with the degree of water recovery and system design. Some relatively small molecules, such as carbon dioxide, hydrogen sulﬁde, silica, and boric acid, may penetrate and reduce water quality. Silica and CO2 are not a problem; low acidity in the product is preferred as a means for dissolving lime in order to add calcium carbonate to the water produced and reduce water aggressiveness. Secondary or higher membrane stages, aeration or ion exchange may solve other problems. The boron problem still exists due to the low rejection of boric acid through the membranes, yet several other solutions exist, as described below. Final mixing of the water is advisable in some cases to increase salt concentration slightly. Small organic compounds dissolved in the feed water may also ﬁnd their way into the water produced. Salt content depends on feed quality (brackish or seawater) and may vary between 50–600 ppm of TDS. A secondary stage may improve quality with only a

j225

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

226

slight cost increase. This is useful in cases where high recovery from seawater is required or where ultrapure water is needed.

10.3 Reverse-Osmosis Desalination: Process and Costs

Figure 10.3 depicts a schematic ﬂow sheet of a typical desalination plant. Feed pretreatment for the removal of suspended material, bacteria, and organics is carried out by sand ﬁltration followed by media ﬁltration. UF or MF modules are used in modern plants. Residual chlorine, if present, is removed with active carbon ﬁlters or by the injection of sodium bisulfate solution. The high-pressure pump used to feed the membrane module may be connected along a single shaft with a motor and a turbine [16, 17] in order to recover the energy content of the pressurized concentrate. Other means, such as independent turbines for secondary stages, may also be used for energy recovery [15]. Concentrate disposal is simple in the case of seawater desalination but more difﬁcult in the case of inland desalination. Measures that can be taken in this case include natural and enhanced evaporation ponds, underground injection, and pipe transport to the sea. The reverse-osmosis membrane process is considered universally as the most promising technology for brackish and seawater desalination [18]. Potential directions for reducing desalination costs may be deduced by analyzing the cost of the components. After the investment, energy is the second cost component to consider. The cost of energy was reduced in the design of the Ashkelon plant with the use of a dedicated gas-turbine power station; this power station reduces energy costs because it is insensitive to the common sine wave of power consumption curve involving ﬂuctuations in day–night, summer–winter electricity demand. Modern energy conservation devices also reduce energy costs, albeit at the expense of increased capital cost. A trend towards increased investment to replace energy will increase with energy cost. Figure 10.4 presents an estimated cost breakdown of desalinated water produced in a typical plant. The main component is, of course, the capital and ﬁnancial cost, comprised of the cost of the main equipment items: feed tanks, pretreatment ﬁltration units, pumps, pressure exchangers and piping, controls, membranes and membranes housing, post-treatment and product tanks. It is obvious from the data in Figure 10.4 that cost reduction may be examined in two main ways. The ﬁrst is to reduce energy cost and the second is to reduce investment. Energy cost depends on the market costs of energy, which are currently rising, and on the efﬁcient use of energy in the process. This is explained later. Investment expenses highly depend on the process operating pressure. The improvement of other parameters will not have a great effect since their consumption is relatively low (manpower, chemicals, membrane replacements, etc.). Good practice, namely good pretreatment, will save membrane replacements. Some items, such as membranes or high-pressure pumps, are restricted to the desalination industry and their cost may simply be lowered by market forces. Investment in sophisticated automation and control equipment can

10.3 Reverse-Osmosis Desalination: Process and Costs

Figure 10.4 Cost estimate of a common RO seawater desalination plant.

reduce water costs by maintaining stable high throughputs and savings in labor costs. As can be seen in Figure 10.4, labor costs are no longer a signiﬁcant cost item since modern desalination plants can operate largely unattended. Wilf [19] presents the energy demand components in a two-pass RO desalination plant. Information about the Ashkelon plant costing may be found in Kronenberg [20] and Velter [21]. More information about RO costing history can be found in Glueckstern [22]. Better predictions for the future are problematic due to the current energy crisis. Compliance with proper operational procedures and following a careful maintenance program can also reduce desalination costs by minimizing the replacement of damaged membranes, reducing the use of cleaning chemicals, and reducing the inventory of membranes and spare parts. The design of a desalination plant is usually a site-speciﬁc task. Pretreatment is the most important local design. It is also envisaged that when operators are insufﬁciently trained, the design and investment will be based invariably on exaggerated safety factors. Well-trained and experienced operators can increase desalination plant production by identifying and debugging bottlenecks. The relatively high cost of seawater desalination can be tolerated easily by the urban customer. In some cases, customers in large cities are paying up to three Euros per m3 of treated water. The monthly cost of water in organized cities is usually lower in comparison to all other utilities. Different industries can usually handle the cost of desalinated water. Some industries need the high quality obtained by desalination; others may reuse and circulate the processed water. The problem is usually in agriculture – simple ﬂood irrigation cannot afford desalination costs. The cost of water in greenhouses is only a small part of the total production cost, allowing for desalination costs. The need for water in agriculture and its cost may be tolerated by ensuring better use of the water. For example, drip irrigation directly to plant roots may save between 50–90% of the water currently used.

j227

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

228

10.3.1 Quality of Desalinated Water

The quality of the water produced can be tailored to meet the needs of the consumer. Practically, it is possible to clean water at a low sodium chloride level. This may be mixed with brackish water or allowed to pass through a bed of CaCO3 to dissolve this salt into the water. The quality of water produced also depends on the quality of brackish water available for mixing. The expected quality of desalinated seawater depends more and more on the permissible concentration level. Seawater contains approximately 5 ppm of boron. Due to the insufﬁcient rejection of current membranes, the water product may contain over 1 ppm of boron. Boron is an important component, especially for plant growth. However, for many crops, too high a boron concentration is harmful and can cause a signiﬁcant reduction in crop yield. Boron may be removed from water by ion exchange, together with secondary and higher RO stages, by increasing the pH of water on the feed side of the membrane and by using EDR. A combination of techniques is also possible [23, 24]. Current demand in Israel requires the production of water containing less than 0.4 ppm of boron in the Ashkelon plant and 0.3 ppm in the future Hadera plant (under design). The reason behind this is related to the recovery of wastewater following the treatment of desalinated water. Boron reaches the wastewater from different sources, which may damage crops irrigated with treated wastewater. In Ashkelon, for example, this demand required using up to four stages of RO membranes to remove the boron, resulting in a signiﬁcant reduction in salts, to a level below 60 ppm TDS. Thermal processes may produce water containing between 5–50 ppm of TDS, similar in composition to feed seawater with thermal techniques. The boron problem does not exist in evaporation techniques. The RO product of brackish water may contain between 200–500 ppm of TDS, which is basically NaCl, and a smaller portion of other salts. Some minor constituents, such as boric acid, hydrogen sulﬁde and CO2, may also be present in the product depending on the composition of the feed water, but may be removed with adequate pre- or post-treatment. Feed water containing dissolved volatile organic compounds will generate water, unless special care is taken, that is slightly contaminated with the same components. This may be true for RO and evaporation techniques. The water product is aggressive, tends to corrode iron pipes, and dissolves protective layers containing calcium and other salts on the inner sides of the mains. This may cause a phenomenon called red water, the release of corrosion products by water dissolving the pipes protective CaCO3 layer. Therefore, water requires posttreatment that usually involves an increase in pH level, the addition of Ca (preferably to a level of about 100 ppm as CaCO3), and alkalinity, namely HCO3 (also to a level of about 100 ppm as CaCO3), according to local water regulations or WHO recommendations. Desalinated water contains a low concentration of salts. Some salts are needed for maintaining a proper balance in bodily functions, so a complementary source of salts is needed for both human and animal diets. Certain agricultural crops may also suffer

10.3 Reverse-Osmosis Desalination: Process and Costs

from a lack of minor constituents when irrigated with desalinated water; others may beneﬁt from this. The addition of magnesium to the desalinated water produced is now being considered for human health and agricultural needs. 10.3.2 Environmental Aspects

Desalination processes may be characterized by their efﬂuents emitted to the environment, the land and atmosphere nearby, and the sea. Desalination is highly dependent on energy, and generally uses fossil energy. All types of air pollution associated with energy production, namely, the emission of NOx, SO2, volatile compounds, particulates and CO2, also exist through the use of electricity produced by conventional power stations or by a dedicated power station. Using gas turbines may increase efﬁciency and therefore reduce pollutants. Efﬂuents from desalination plants contain a relatively high concentration of salts and depend on water recovery from the feed brine. In the case of seawater desalination, rejected brine is almost twice as concentrated as the original seawater solution. The concentrate also contains chemicals used in the pretreatment of the feed water. The latter may contain low concentrations of antiscalants, surfactants and acid added to the feed water that reduces pH. Occasional washing solutions or rejected backwash slurries from feed water may be added to this. In small-scale operations, the problem is minor and no serious damage is caused to marine life. In large-scale water production, the problem is more serious. However, the dilution and spreading of efﬂuents may solve this problem. Natural chemicals that do not harm the environment will probably replace part of the added chemicals in the future. Concentrates should be released to the sea a few meters above the sea ﬂoor from a few nozzles pointing upward, at an angle between 45–60 degrees. The volume of high concentration will be minimized, no damage will be caused to marine vegetation and small species, and ﬁsh will avoid this region. The more serious problems involve concentrates that are produced in-land in the case of brackish-water desalination or wastewater recovery. The concentrate composition here differs from the seawater composition. In most cases, the solution contains more calcium and magnesium; sometimes other components are involved depending on the composition at the source. The problem is less severe when the solutions are purged into the open sea; in this case, care must be taken to prevent possible salt precipitation from supersaturated concentrates along the discharge pipeline [25]. If the sea is not easily accessible, the concentrate may increase groundwater salinity if allowed to penetrate into the ground. A possible solution to this problem involves zero discharge treatment, which is the evaporative separation between solids and water to enable solids to be stored properly in-land. This solution may be performed using solar evaporation ponds or by forced evaporation with available heat sources. The process is expensive, but the basis for comparison is the cost of transporting brine to the nearest possible authorized area, taking into account the inﬂuence of this treatment on product cost. Another approach is adding a crystallizer working at elevated pH to remove the supersaturation of the dissolved

j229

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

230

salts from the concentrate stream. In most cases, it is then possible to add a second RO stage to produce more water [26]. 10.3.3 Energy Issues

Desalination, as a separation process, requires energy. The speciﬁc energy for reverse-osmosis desalination has decreased signiﬁcantly over the past decade and is approaching the theoretical thermodynamic minimum. This was achieved through the development of large pumps having an efﬁciency as high as 92% equipped with modern, efﬁcient turbines and other energy-recovery devices. The newer devices, known as turbochargers, pressure exchangers or work exchangers (names adopted by different producers), represent efﬁcient ways of recovering the energy content of the high-pressure concentrate. Turbines turn concentrate pressure into the velocity of jets that spin a wheel, which is used either to reduce the power consumption of the motor driving the pump or to boost the pressure of the feed to a second stage. Other methods of exchanging the pressure of the exiting concentrate involve simple devices transferring pressure to the seawater feed. Using the new techniques, a reduction in power consumption of the desalination systems was achieved. For example, processing 3.5% seawater at a recovery rate of 50% requires about 2.7 KWh/m3 of the water produced by recovering the concentrate energy using turbines. Pressure exchangers can go even lower, to about 2.2 kWh/m3 of water produced. Since more energy is consumed for feed and concentrate pumping, as well as for the pretreatment stages, overall energy needs are less than 3.7 kWh/m3 for seawater production. Production from brackish water, wastewater or slightly polluted sources can go as low as 1.7 kWh/m3 of water produced or less, depending on salts concentration and possible recovery. The energy cost of an optimized desalination plant is approximately 30–40% of the total water cost. This cost may vary since energy costs may be replaced by equipment investment. For example, fossil energy may be replaced theoretically by different types of solar collectors. The problem is that since solar energy is available only 25% of the time, the investment fraction increases by a factor of four, before taking into account the equipment needed for electricity production. The water cost will be much higher compared to the use of a regular energy source. The optimization is made during the design of the plant, yet the energy cost may vary signiﬁcantly during the projects lifetime. For example, during the writing of this chapter, the cost of natural oil was increasing at a signiﬁcant rate in comparison to its cost at the design stage of the Ashkelon plant. It is difﬁcult to change the optimal design of a plant once it has been built. However, it is possible to minimize losses by designing for more ﬂexible changes in terms of variable energy consumption and equipment costs. A 100-million-m3 RO-based seawater desalination plant requires an electrical energy supply of less than 50 MW. A dedicated power station can work at a much higher efﬁciency than a regular power station for this purpose since it is operated constantly without the known sine wave, representing day–night, summer–winter

10.3 Reverse-Osmosis Desalination: Process and Costs Table 10.1 RO energy consumption in comparison to other alternatives.

Fuel Subject

Natural gas

Gasoil

Heavy fuel

Coal

Caloric value kcal/kg fuel Caloric Value kWh/kg fuel Electricity production (45% eff.) kWh/kg fuel large power station Electricity production (80% eff.) high-efﬁciency gas turbine kWh/kg fuel Capacity – seawater desal. (50% recovery) m3/kg fuel 80% efﬁciency Fuel consumption/ton desalinated water kg fuel/m3 80% efﬁciency How many km can I drive with 1 m3 desalinated water fuel consumption? How many hours of a single room AC (2.5 kWh) can I operate?

9000 10.5 4.7

10 750 12.5 5.6

10 000 11.6 5.2

7700 9 4

1.3

1.6

1.5

1.2

2.4 0.7

0.6

0.7

0.9

0.4 2–7

2–6

8.4

1.4

changes in consumption. Better efﬁciency is expected for gas turbines since the high temperature of the gases may also be used. Therefore, the real energy required is lower than for other common uses. Critics among environmentalists often express concern about the energy consumption for water desalination. Water is needed for many people on earth and for supplying the basic needs of the majority of these people. The introduction of desalination may leave less water for the environment. This takes priority over using energy for air-conditioning or running large, energy-consuming cars. A look at Table 10.1 will show the real energy consumption and a comparison to other energy usages. The table shows consumption in terms of different sources of energy, natural gas, gasoil, heavy fuel and coal. This is the best way of comparing energy demand when comparing electricity and fuel. A large-scale power station generates electricity from coal or heavy fuel at around 45% efﬁciency. If operated effectively, gas turbines may go as high as 80% efﬁciency. Fuel consumption, in terms of how many kWh is produced from 1 kg of fuel, is presented in the table. Next, follows the calculations of how many cubic meters (tons) of water can be produced from a single kg of fuel by using a large RO plant. From here we can see how much fuel is needed to make a ton of water, the cheapest product on earth. At this point, one can see that this amount of energy can drive us only 2–7 kilometers in our car (depending on its size), or determine how long we can operate our small room AC with the same amount of fuel. Looking at these numbers, the cost of energy for desalination in comparison with other energy usages is rather small. Environmental concern about the CO2 greenhouse effect associated with the use of hydrocarbon fuel has led to the goal of supplying desalination energy from

j231

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

232

renewable-energy sources. Renewable-energy sources may soon be compatible and economic for general electricity production. At this stage, they will also be suitable for desalination purposes. No doubt, greater efforts should be devoted towards exploiting renewable-energy sources. However, the real test of any new energy source is its acceptance for electricity production or other common energy uses. Savings on CO2 emissions must be made in terms of other energy forms and not regarding the very delicate issue of desalination for freshwater production. The use of nuclear energy, which is currently more expensive than fossil energy, is dangerous in areas where political instabilities prevail. It is also problematic where the technology is not accessible and it is necessary to rely on imported, trained, and sophisticated labor. A possible method of efﬁcient energy use in a sufﬁciently large desalination plant involves the design of a hybrid plant consisting of a membrane unit combined with a vapor-compression unit [27] using electrical energy and a multieffect evaporation plant using heat energy. Such an operation is common in the chemical industry. Energy costs could be minimized by coupling the desalination plant with a dedicated power plant generating electricity and waste heat at optimal economic conditions. The advantage of the day–night, summer–winter electricity production cycle is that desalinated water is produced during the night, involving lower power consumption. The main disadvantage is that the desalination equipment is not used for a large percentage of the time. This is a mistake, since, as in any modern plant, production costs are greater if the equipment is not in full use. An efﬁcient desalination plant should therefore be operated 24-hours-a-day, 365-days-a-year, with exceptions only for maintenance. During this time, a full supply of energy is required at the lowest cost.

10.4 Treatment of Sewage and Polluted Water

A large source of water for reuse may be obtained by reclaiming polluted water. Sources of polluted water emanate from domestic wastewater, industrial waste solutions, agricultural efﬂuent as runoff water, recirculated greenhouse water, and ﬁsh pond waste. All these must be treated to a tolerable quality to prevent deterioration of the soil and aquifers, and pollution of lakes, rivers, and the sea. Above all, this is a source of usable water. The current global trend in dealing with this important problem involves secondary biotreatment. This is followed by the increased use of membranes. MF and UF membranes are capable of almost completely removing suspended matter, including bacteria 6–9 orders of magnitude (6–9 logs of removal), waterborne protozoa, and reducing virus content by two to three orders of magnitude. Parameters affecting virus removal are associated with particulate agglomeration occurring next to the membrane, possible adsorption to the cake layer, virus association in groups, poresize reduction by the cake, gel layer and fouling build-up on the membrane, and also pH effects. RO or NF membranes can be used later to remove salinity and some

10.4 Treatment of Sewage and Polluted Water

Figure 10.5 Schematic presentation of a wastewater treatment plant.

dissolved organic matter, while reducing TOC, COD, and BOD. Water can be disinfected by UV, ozone, chlorine, and chloramines. Figure 10.5 illustrates the wastewater-treatment process. It is possible to use strained, presettled wastewater primary efﬂuent or secondary efﬂuent following biological treatment. Proper usage of the biotreatment removes the organic matter and almost all ammonia and phosphates from water solutions. Anaerobic operation reduces nitrates to elementary nitrogen. When membranes are submerged in a bioreactor, it is known as a membrane bioreactor (MBR) [11]. Efﬂuent from the bioreactor may also be treated with external MF or UF membranes, and the concentrate is circulated back to the bioreactor. Water at this stage may be used mainly for irrigation also in some places but as drinking water. The MBR-treated efﬂuents can be fed directly into the RO/NF system. The ﬁnal product, following RO/NF, surpasses current (and future) environmental requirements by far, as well as those for unrestricted use in agriculture, aquifer reinﬁltration, and eventually evens all municipal uses. Polishing with AC to remove trace organics may be used as means of reducing RO/ NF membrane fouling and deterioration and ﬁnal polishing following RO/NF product. Water exiting the last membranes requires the addition of calcium and possibly magnesium salts to be accepted by WHO recommendations. Certain parameters affect membrane fouling: particle nature; particle size and size distribution; membrane type and structure; surface interactions; and the clogging mechanism. An important parameter is the method applied to the ﬁltration technique, namely, crossﬂow or deadend ﬁltration. The latter requires less pumping energy but tends to clog the membrane faster.

j233

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

234

10.4.1 Membrane Bioreactors

Combining UF or MF membrane technologies with biological reactors for the treatment of wastewater in a one-stage process has led to the generation of the MBR concept in which MF or UF membranes have replaced the traditional sedimentation tank. An efﬁcient clariﬁcation of the treated wastewater is achieved, membranes can reduce the disinfection practices such as chlorination, and a pathogen-free, tertiary quality efﬂuent is thus obtained [11]. The main advantages of MBR over waste activated sludge processes are: . . . . . . .

Small footprint; Complete solid removal; Efﬂuent disinfection; High loading rate capabilities; Low to zero sludge production; No bulking problems; High oxygen utilization rate. Other parameters that require attention are:

. . . . . . .

Operation with deadend or crossﬂow; Membrane conﬁguration; Cleaning of fouled membranes; Washing; Frequency of backwashing; Use of air; Use of cleaning solutions.

Different sources of wastewater may contain different materials, so every type of wastewater must be tested in order to choose the right conditions. Harussi et al. [28] compared the alternative costs of feed-water desalination for cities to wastewater desalination. The ﬁrst alternative may be cheaper but the second is necessary in order to be compatible with the environment. 10.4.2 Reclaimed Wastewater Product Quality

Product quality depends heavily on previous treatment stages. Organic matter and ammonia may exist in the product in the cases of insufﬁcient removal. However, low ammonia contentmayreact with chlorine toform chloramines,a long-termdisinfector. Chlorination of organic matter may result in halogenated organic matter. Insufﬁcient removal of phosphates in the biotreatment stage will result in a reduced recovery of the RO process if supersaturation is reached [29]. Insufﬁcient de-nitriﬁcation, the removal of nitrates, will cause high nitrate concentration in the concentrate, which is a problem with concentrate removal. Boron content increases in the city due to many sources, especially from detergents and from industries using bleaching processes.

10.5 Fouling and Prevention

The concentration of urban wastewater contaminants is relatively low, so highrecovery membrane processes can generally be used to solve the problem relatively inexpensively. The case is completely different regarding industrial wastewater; each stream having different contaminants must be dealt with independently. Many treatment facilities at different locations were installed to produce water from wastewater for different uses. In some cases, MF membranes are used directly on strained wastewater to remove suspended particles that are too large for the gap between two membranes [30]. Simple wastewater-treatment facilities in Europe exist along all large rivers. Secondary treated waters ﬂowing into the rivers are again pumped at a distance of about 200 meters downstream, treated with active carbon and UF membranes, disinfected and then distributed to the system. This is wastewater treatment without an RO section due to the low salinity of the water. The process cannot handle dissolved medicines, hormones, drugs, and other contaminants that could be removed with RO or NF membranes. In some cases, NF membranes are used for better treatment of the water. Information on wastewater costing may be found in Adham et al. [31].

10.5 Fouling and Prevention

The main problem in membrane usage for water puriﬁcation is the fouling layer that adheres to the membrane. The source of the fouling layer is the different species existing in the feed water and their increased concentration next to the membrane wall. When water permeates through the membrane, all rejected species accumulate next to the membrane wall, their concentration increases in comparison to the bulk concentration, and the motion away from the membrane is controlled by diffusion to the bulk of ﬂow against the ﬂux of the water ﬂowing to the membrane. The main types of fouling are: 1. 2. 3. 4.

Suspended particles. Salt precipitation due to supersaturation. Dissolved organic matter. Biofouling.

The denser RO and NF membranes may face all of the above problems. UF and MF are open membranes and hence encounter problems associated with salt precipitation. Other important parameters are listed below: . . . . . .

Clogging by suspended solids, precipitating salts, bacteria cultures, and so on. Difﬁculties in membrane surface cleaning. Sensitivity to degradation by organic chemicals. Sensitivity to different types of bacteria. Deterioration due to free chlorine. Permeability of nonpolar substances, including low molecular weight organic substances.

j235

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

236

The critical issue for a successful RO plant is pretreatment. Long-term operating experience proves the viability of continuous MF/UF pretreatment of RO for the desalination of a wide variety of water sources. MF/UF has proven to simplify and reduce the costs of traditional pretreatment, comprised of deep-bed media ﬁlters combined with chemical treatment. MF/UF produces ﬁltrate of a consistent quality almost irrespective of ﬂuctuations in feed-water quality. In the last ﬁve years, ROmembrane improvements, combined with the use of membrane ﬁltration for pretreatment, have halved the cost of advanced treatment and are now more widely used for the reuse of municipal wastewater. Suspended-particle precipitation is caused by attraction forces with the membrane wall. They may be removed effectively by good pretreatment, either by sand ﬁltration or by using UF/NF membranes. Salt precipitation is caused by the increased concentration of low solubility salts such as calcium salts – carbonate, ﬂuoride, and sulfate (mainly from brackish water), phosphate (domestic wastewater), silica, iron oxides, and other metal salts originating from water feed. Many techniques are available to prevent their precipitation, such as acidulation to remove carbonates or increase the solubility of salts, the removal of special ions by ion exchange, hardness removal prior to feeding the membranes, and so on. This is not a major problem in seawater desalination, however, it is very important in brackish water and recovered wastewater desalination. The recovery level of water produced from a certain source is controlled by the ability to maintain high supersaturation of the salts before precipitation. This is done by adding antiscalants or crystallization inhibitors [32]. The antiscalants are usually medium-length polymers, such as polyphosphates, polyphosphonates, and polyacrylates. Their mechanism is not completely understood, yet they interfere with crystal growth through adsorption on the active sites or by applying electrical charges that prevent the crystals from growing. Techniques were developed to estimate the recovery level while comparing different antiscalants at different concentrations [33–39]. Organic fouling is the basis of biofouling, which is one of the most severe problems seeking a solution [40–43]. Currently, only good pretreatment may limit biofouling. 10.5.1 How to Prevent

Like in medicine, prevention is the key in most cases for healthy membrane life. A good pretreatment allows much longer membrane life at lower cleaning costs. Pretreatment is used mainly for the removal of suspended solids, bacteria, and large organic molecules. This is done mostly by sand ﬁltration, but UF membranes can also be found for this purpose in modern designs. Dosing the water with coagulants is used to agglomerate the small particles for easy separation. dissolved air ﬂotation (DAF) was also introduced as a means for removing both suspended and organic matter [44, 45]. Disinfection of water is used to destroy bacteria. However, RO membranes are sensitive to oxidizers so they must be removed before entering the membranes. Active-carbon beds are sometimes used to remove traces of organic matter together

10.6 R&D Directions

with excess chlorine. It was found in some cases that improper usage brings the bacteria level, after the active carbon, back to the original count before treatment. Other pretreatment steps involved in the removal of special contaminants include acidulation, ion exchange of some important contaminants, and the removal of H2S, if it exists. 10.5.2 Membrane Cleaning

Membrane cleaning is an important stage that should be minimized, if possible. Cleaning fouled membranes depends on the type of fouling. A simple wash along the membrane or opposite the ﬂow direction is the easiest type. Low pH is used when salts are precipitated. High pH is effective when silica is precipitated, and includes organic matter or even bacteria. Backwash is the usual treatment of UF/NF membranes, usually involving the dosing of disinfectant. It is impossible, however, to apply high pressure to the permeate side in RO/NF operations since this may damage the membranes. Recent developments have shown that backwash is possible in these membranes based on direct osmosis, the difference between osmotic pressure of the feed side and the permeate side. Short, frequent osmotic backwash may maintain clean membranes and increase their life. Instructions for cleaning methods are given by membrane manufacturers.

10.6 R&D Directions 10.6.1 Impending Water Scarcity

The need for more, better-quality water is increasing all the time. Global warming, either naturally or as a result of excessive use of fossil energy, is causing glaciers to melt and consequently ocean water levels to rise. This alone will cause signiﬁcant problems associated with current water systems. River levels will rise, ﬂooding their immediate surroundings. An increase in seawater level will change the current balance between seas and shore aquifers, causing a major penetration of seawater into the aquifers. There is no way to accommodate these problems other than by relying on desalinated water and wastewater recovery. Additional R&D work is therefore required in order to continue to reduce the cost of water by implementing improved recovery techniques and ensuring better usage in agriculture. 10.6.2 Better Membranes

The main components of RO desalination costs are energy cost and equipment investment. Itemizing the equipment in use, membranes, pressure vessels, pumps,

j237

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

238

tubing, and ﬂow devices and energy-recovery units illustrate that there is no special item that is signiﬁcantly more expensive than the others. However, membranes play the most important role in possible cost reduction. The cost of RO membranes represents about 8% of the overall investment. Membranes may be improved signiﬁcantly. Permeation may be increased, maintaining similar rejection properties. Increased ﬂux through the membranes will enable a pressure reduction and hence less energy at the same recovery ratio. The pressure reduction may also reduce the costs of the expensive metals used at high pressure in a highly corrosive environment. This will reduce the cost of pumps and ﬂow devices. Larger membrane modules will reduce the plants footprint. Other important future membranes properties include improved resistivity to extreme pH enabling better cleaning performance and resistance to oxidizers, organic solvents and particulate fouling. More important properties exist, yet the most important may be the resistance to fouling of the different sources. Similar properties are needed for other types of membranes in use such as MF, UF, and NF membranes. Improved membranes may be used in other separation processes, not necessarily related to water [46–48]. 10.6.3 New Membranes-Based Desalination Processes

Recent developments have brought signiﬁcant attention to other types of membrane processes reported mainly in research papers: forward osmosis [49, 50], and membrane distillation [51, 52]. Forward osmosis, or direct osmosis, is deﬁned by water passage from a salt solution or a polluted solution through a membrane to a solution containing dissolved matter of higher osmotic pressure. A possible advantage of such a process is that the separation of the water from the higher osmotic pressure solution is easier than the separation through RO. These separations were proposed by using magnetic nanoparticles covered with organic matter, separated by magnetic ﬁeld, distillation of the dissolved material like in the case of ammonium carbonates, or a possible simpler separation, such as crystallization. It is also important to minimize contaminant traces of the high osmotic-pressure material in product water according to drinking-water regulations. Thermodynamics, however, teaches us that minimum separation energy is dependent on concentration. RO separation is very close to minimal thermodynamic separation energy. So, the objective here is to ﬁnd a process that consumes less energy than the RO process. Another new trend is called membrane distillation. This is based on open hydrophobic membranes that enable the passage of water vapor only. The product quality is expected to be better than RO since only water vapor may pass through the membrane. Vapor condensation is allowed on colder surfaces adjacent to the membranes or outside the membrane module, where vapors are pumped out. Another way is to condense the vapor in direct contact with a cold-water stream. The main problem using this technique is the need to evaporate the water. The energy demand for this is around 650 kWh/m3. This enormous amount of energy may be reduced when energy reuse is possible, in a similar way to the multieffect distillation

10.6 R&D Directions

desalination process. This may reduce the energy demand down to about 60 kWh/m3 if energy is reused more than 10 times within the desalination plant. More energy is needed for pumping the water and for cooling. Heat-transfer ﬂux through the membrane is low, so a large transfer area is needed. Developers of this technique claim that the high energy demand may be supplied by low-grade, cheap energy yet this claim is also true for the multieffect distillation process. The only possible advantage of the multieffect distillation technique is the possibility of lower volume and low footprints of the plant. However, the design is complicated, as is shown in Figure 10.6, representing the preliminary design of membrane distillation with heat recovery, similar but not identical to the multieffect distillation process. In this design, entering seawater exchanges heat, leaving hot streams of product and concentrate. External heat must be added to the feed water. The water is then allowed to ﬂow in parallel to the membranes where evaporation takes place. Vapor leaving the membrane condenses on the heat-transfer wall and transfers the heat of condensation to the water in the next stage. In this way, the evaporation energy is reused to heat the concentrating solution. Other designs are possible, yet in terms of energy demands, the technique cannot compete with RO.

Figure 10.6 Schematic view of a membrane distillation design based on multieffect distillation technology.

j239

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

240

10.7 Summary

The use of different types of membranes for water production and puriﬁcation is presented here. Solutions are available for good-quality water production at affordable costs. Important related aspects such as environmental and energy issues are presented. Future directions are reviewed. Additional research is required in order to improve the processes and reduce the cost of water produced. The subject of water is one of the most important subjects that humankind must solve, together with renewable energy and environmental problems.

References 1 Semiat, R. (2000) Desalination – Present and future, Invited Article for IWRA21. Water International, 25 (1), 54–65. 2 Faller, K.A., Murray, P. and Livingston, A. (1998) Reverse osmosis and nanoﬁltration manual of water supply practices, AWWA Manual of Water Supply Practice, M46 Amer Water Works Assn., Elbert, Colorado, USA. 3 Wilbert, M.C., Leitz, F., Abart, E., Boegli, B. and Linton, K. (1998) The Desalting and Water Treatment Membrane Manual: A Guide to Membranes for Municipal Water Treatment, 2nd edn, Water Treatment Technology Program Report No. 29, Bureau of Reclamation, United States Department of Interior. 4 Bergman, R.A.(Nov. 1995) Florida – A Cost Comparison Update, Membrane Softening vs. Lime Softening. International Desalination and Water Reuse. 5 Hassan, A.M., Al-Soﬁ, M.A.K., Al-Amoudi, A., Jamaluddin, T.M., Dalvi, A.G.I., Kitner, N.M., Mustafa, G.M. and Al-Tisan, I.A. (1998) A new approach to membranes and thermal seawater desalination processes using nanoﬁltration membranes. International Desalination and Water Reuse Part 1: May 1998, Part 2: Aug. 6 Sagiv, A. and Semiat, R. (2005) Backwash of RO spiral wound membranes. Desalination, 179, 1–9. 7 Liberman, B. (2004) Methods of direct osmosis membrane cleaning online for

8

9

10

11

12

13

14

high SDI feed after pre-treatment. IDA Workshop, Tempa-San Diego. Altman, M., Hasson, D. and Semiat, R. (1999) Dynamic membranes. Reviews in Chemical Engineering, 15, 1–40. Rumyantsev, M., Shauly, A., Yiantsios, S.G., Hasson, D., Karabelas, A.J. and Semiat, R. (2000) Parameters affecting the properties of dynamic membranes formed by Zr hydroxide colloids. Desalination, 131, 189–200. Baudin, I., Chevalier, M.R., Anselme, C., Cornu, S. and Laine, J.M. (1997) LApie and Vigneux Case studies: ﬁrst months of operation. Desalination, 113, 273–275. Yang, W.B., Cicek, N. and Ilg, J. (2006) State-of-the-art of membrane bioreactors: Worldwide research and commercial applications in North America. Journal of Membrane Science, 270, 201–211. Hai, O.H. and de Ryck, L.(Sept. 2005) Water reuse and Ulu Pandan NEWater project. IDA World Congress on Desalination and Water Reuse, Singapore. Thampy, S.K., Rangarajan, R. and Indusekhar, V.K.(August 1999) 25 years of electrodialysis experience at Central Salt and Marine Chemicals Research Institute Bhavnagar, India. International Desalination and Water Reuse, Vol. 9/2. Strathmann, H. (2004) Assessment of electrodialysis water desalination process costs. Proceedings of the International

References

15

16

17

18

19

20

21

22

23

24

25

Conference on Desalination Costing, Limassol, Cyprus. Voutchkov, N. and Semiat, R. (2008) in Advanced Membrane Technology and Applications (eds N.N. Li, W.S. Winston Ho and A.G. Fane), John Wiley & Sons, Inc., New York. Glueckstern, P. and Priel, M. (1999) Experience, capability and plans for erecting large seawater desalination plants. Proceedings of the 2nd Annual IDS Conference, Haifa. Glueckstern, P. and Priel, M. (2000) Desalination of brackish and marginal water sources in Israel: Past present and future. Proceedings of the 3rd Annual IDS Conference, Tel-Aviv. Furukawa, D.H.(Sept. 1997) A review of seawater reverse osmosis. IDA Desalination Seminar, Cairo, Egypt. Wilf, M. (2004) Fundamentals of RO-NF Technology. Proceedings of the International Conference on Desalination Costing, Limassol, Cyprus. Kronenberg, G. (2004) Ashkelon 100 MC/ Year BOT Project. Proceedings of the International Conference on Desalination Costing, Limassol, Cyprus. Velter, G. (2004) Ashkelon desalination project operations. Proceedings of the International Conference on Desalination Costing, Limassol, Cyprus. Glueckstern, P. (2004) History of desalination cost estimations. Proceedings of the International Conference on Desalination Costing, Limassol, Cyprus. Glueckstern, P. and Priel, M. (2003) Optimization of boron removal in old and new SWRO systems. Desalination, 156, 219–228. Sagiv, A. and Semiat, R. (2004) Analysis of parameters affecting boron permeation through reverse osmosis membranes. Journal of Membrane Science, 243 (1–2), 79–87. Semiat, R., Hasson, D., Zelmanov, G. and Hemo, I. (2004) Threshold scaling limits of RO concentrates ﬂowing in a long waste disposal pipeline. Water

26

27

28

29

30

31

32

33

34

35

Science and Technology, 49 (2), 211–219. Lisitzin, D., Hasson, D. and Semiat, R.(June 2006) Membrane crystallizer for increased desalination recovery, ECI – Advanced Membranes Technology III. Membrane Engineering for Process Intensiﬁcation, Cetraro, Calabria, Italy. Awerbuch, L. (1997) Dual purpose power desalination/hybrid systems/energy and economics. IDA Desalination Seminar, Cairo, Egypt. Harussi, Y., Rom, D., Galil, N. and Semiat, R. (2001) Evaluation of membrane processes to reduce the salinity of reclaimed wastewater. Desalination, 137, 71–89. Greenberg, G., Hasson, D. and Semiat, R. (2005) Limits of RO recovery imposed by calcium phosphate precipitation. Desalination, 183, 273–288. Johnson, W.T., Phelps, R.W. and Beatson, P.J. (Oct. 1997) Wastewater reuse using membranes. IDA World Congress on Desalination and Water Science, Madrid. Adham, S., Kumar, M. and Pearce, W.H. (2004) Development of a model for brackish and reclaimed water membrane desalination costs. Proceedings of the International Conference on Desalination Costing, Limassol, Cyprus. Hasson, D. and Semiat, R. (2006) Scale control in saline and wastewater desalination. Israel Journal of Chemistry, 46 (1), 97–104. Hasson, D., Semiat, R., Bramson, D., Busch, M. and Limoni-Relis, B. (1998) Suppression of CaCO3 scale deposition by anti-scalants. Desalination, 118, 285–296. Drak, A., Glucina, K., Busch, M., Hasson, D., Laine, J.M. and Semiat, R. (2000) Laboratory technique for predicting the scaling propensity of RO feed waters. Desalination, 132, 233–242. Hasson, D., Drak, A. and Semiat, R. (2001) Inception of CaSO4 scaling on RO membranes at various water recovery levels. Desalination, 139, 73–81.

j241

j 10 Seawater and Brackish-Water Desalination with Membrane Operations

242

36 Hasson, D., Drak, A. and Semiat, R. (2003) Induction times induced in an RO system by antiscalants delaying CaSO4 precipitation. Desalination, 157, 193–207. 37 Semiat, R., Sutzcover, I. and Hasson, D. (2001) Technique for evaluating silica scaling and its inhibition in RO desalting. Desalination, 140, 181–193. 38 Semiat, R., Sutzkover, I. and Hasson, H. (2003) Characterization of the effectiveness of silica anti-scalants. Desalination, 159, 11–19. 39 Semiat, R., Sutzkover, I. and Hasson, D. (2003) Scaling of RO membranes from silica supersaturated solutions. Desalination, 157, 169–191. 40 Ivnitsky, H., Katz, I., Minz, D., Shimoni, E., Chen, Y., Tarchitzky, J., Semiat, R. and Dosoretz, C.G. (2005) Characterization of membrane biofouling in nanoﬁltration processes of wastewater treatment. Desalination, 185, 255–268. 41 Kang, S., Asatekin, A., Mayes, A.M. and Elimelech, M. (2007) Protein antifouling mechanisms of PAN UF membranes incorporating PAN-g-PEO additive. Journal of Membrane Science, 298, 42–50. 42 Asatekin, A. et al. (2006) Antifouling nanoﬁltration membranes for membrane bioreactors from self-assembling graft copolymers. Journal of Membrane Science, 285, 81–89. 43 Kimura, K., Yamato, N., Yamamura, H. and Watanabe, Y. (2005) Membrane fouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater. Environmental Science & Technology, 39, 6293–6299. 44 Huehmer, R. and Henthorne, L. (2006) Advances in RO pretreatment techniques innovations and applications of sea-water and marginal water desalination. IDS 8th Annual Conference Technion, Haifa. 45 Bonnelye, V., Sanz, M.A., Mazounie, P., Vion, P., Del Castillo, J. and Rovel, J.M. (2006) Advances in DAF high rate ﬂotation, innovations and applications of sea-water and marginal water desalination.

46

47

48

49

50

51

52

53

54

IDS 8th Annual Conference Technion, Haifa. Akthakul, A., Salinaro, R.F. and Mayes, A.M. (2004) Antifouling polymer membranes with sub-nanometer size selectivity. Macromolecules, 37, 7663–7668. Ulbricht, M. (2006) Advanced functional polymer membranes. Polymer, 47, 2217–2262. Eisen, M. and Semiat, R. (2008) Membranes in Desalination and Water Treatment. To appear in MRS publication, special issue on materials for water. McCutcheon, J.R., McGinnis, R.L. and Elimelech, M. (2005) A novel ammoniacarbon dioxide forward (direct) osmosis desalination process. Desalination, 174, 1–11. Cath, T.Y., Adams, D. and Childress, A.E. (2005) Membrane contactor processes for wastewater reclamation in space II. Combined direct osmosis, osmotic distillation, and membrane distillation for treatment of metabolic wastewater. Journal of Membrane Science, 257 (1–2), 111–119. Alklaibi, A.M. and Lior, N. (2005) Membrane-distillation desalination: Status and potential. Desalination, 171 (2), 111–131. Gilron, J., Song, L. and Sirkar, K.K. (2007) Design for cascade of crossﬂow direct contact membrane distillation. Industrial & Engineering Chemistry Research, 46, 2324–2334. Blanco, J.F., Sublet, J., Nguyen, Q.T. and Schaetzel, P. (2006) Formation and morphology studies of different polysulfones-based membranes made by wet phase inversion process. Journal of Membrane Science, 283, 27–37. Asatekin, A., Kang, S., Elimelech, M. and Mayes, A.M. (2007) Anti-fouling ultraﬁltration membranes containing polyacrylonitrile-graft-poly(ethylene oxide) comb copolymer additives. Journal of Membrane Science, 298, 136–146.

References 55 Hasson, D., Drak, A., Komlos, C., Yang, Q. and Semiat, R. (2007) Detection of fouling on RO modules by residence time distribution analyses. Desalination, 204 (1–3), 132–144. 56 Hester, J.F. and Mayes, A.M. (2002) Design and performance of foul-resistant poly(vinylidene ﬂuoride) membranes prepared in a single step by surface

segregation. Journal of Membrane Science, 202, 119–135. 57 Wang, Y.Q. et al. (2005) Remarkable reduction of irreversible fouling and improvement of the permeation properties of poly(ether sulfone) ultraﬁltration membranes by blending with pluronic F127. Langmuir, 21, 11856–11862.

j243

j245

11 Developments in Membrane Science for Downstream Processing Jo~ao G. Crespo

11.1 Introduction

This chapter discusses the use of membrane processes for recovery, concentration, and puriﬁcation of biologically active compounds from complex media. This chapter is not organized and written as a review paper aiming at referring all major developments in the use of membranes for downstream processing but, rather, it presents the authors perspective about this ﬁeld, its main constraints and challenges. Most scientiﬁc reviews in this ﬁeld [1–3] are focused on the recovery and puriﬁcation of large molecules with biological activity – protein harvesting and protein puriﬁcation (including puriﬁcation of monoclonal antibodies), puriﬁcation of DNA and RNA – but the recovery of small molecules [4] with biological activity (M < 500 Da) usually attracts little attention. This chapter will pay attention to both small and large biomolecules and will discuss the critical issues related with their recovery and puriﬁcation. Membrane bioreactors are excluded from this discussion, although some of the most interesting processes under development involve the integrated concept of bioconversion and product recovery [4–6]. 11.1.1 Why Membranes for Downstream Processing?

Membrane separations are regarded as particularly suitable for biotech applications because (1) in general, they can be operated under mild conditions of temperature, pressure, and shear stress, therefore preserving the biological activity of the compounds to be recovered and the properties of the original media/matrices; (2) they do not require any extraction mass agents such as solvents, avoiding product contamination and the need for subsequent puriﬁcation and (3) a large variety of membrane materials are available: they can be polymers, inorganic matrices or composites. The wide range of possible materials underlines one of the strengths of membrane separations: the possibility of designing and ﬁne tuning the membrane for a speciﬁc

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 11 Developments in Membrane Science for Downstream Processing

246

need, through development of materials with adequate structural properties or by suitable modiﬁcations of their surface chemistry and topography [7–9]. Additionally, membranes have the unique advantage of allowing the simultaneous contact with two different media, at each membrane side, creating compartments with different properties. Therefore, membranes offer the potential to promote the spatial organization of catalytic compartments and selective barriers. This feature is used with advantage in new concepts of membrane multiphasic (bio)reactors and membrane contactors. When using porous membranes for ﬁltration processes, they act primarily on the basis of size exclusion, leading to permeate ﬂuxes that are high when compared with other competing processes, due to convective transport through the porous structure of the membrane. Therefore, high throughputs are usually referred to as one advantage of membrane ﬁltration processes, such as microﬁltration and ultraﬁltration, when compared with other separation processes involving porous media, as happens with chromatographic systems.

11.2 Constraints and Challenges in Downstream Processing

The main constraints and problems associated with the use of membrane processes for downstream processing have been extensively discussed in the literature and the understanding of their nature and mechanisms has driven research towards the development of new solutions. 11.2.1 External Mass-Transport Limitations

Mass-transport limitations are common to all processes involving mass transfer at interfaces, and membranes are not an exception. This problem can be extremely important both for situations where the transport of solvent through the membrane is faster and preferential when compared with the transport of solute(s) – which happens with membrane ﬁltration processes such as microﬁltration and ultraﬁltration – as well as with processes where the ﬂux of solute(s) is preferential, as happens in organophilic pervaporation. In the ﬁrst case, the concentration of solute builds up near the membrane interface, while in the second case a depletion of solute occurs. In both situations the performance of the system is affected negatively: (1) solute accumulation leads, ultimately, to a loss of selectivity for solute rejection, promotes conditions for membrane fouling and local increase of osmotic pressure difference, which impacts on solvent ﬂux; (2) solute depletion at the membrane surface diminishes the driving force for solute transport, which impacts on solute ﬂux and, ultimately, on the overall process selectivity towards the transport of that speciﬁc solute. A large number of methods for improving external mass transport in membrane systems have been proposed and evaluated. Several of them may lead to a signiﬁcant process improvement under deﬁned conditions. Still, these methods – use of

11.2 Constraints and Challenges in Downstream Processing

corrugated membranes [10], module development using Taylor vortices [11] and Dean vortices [12], vibrating modules [13], use of optimized spacers and static promoters [14] – refer to situations where the energy input is not totally used at the correct location, the membrane surface locals where solute buildup or depletion is occurring, being partially lost within the bulk liquid. New strategies, involving the fabrication of membranes with a speciﬁcally designed surface 3D topography have been proposed and developed [15]. Besides their use as nanostructured surfaces for cell differentiation these membranes, fabricated with different techniques such as nanoimprint lithography, open new opportunities for improved mass transport at the membrane scale. 11.2.2 Membrane Fouling

Fouling is the most used word in the membrane literature. Considering the character of most biological media, where proteins and polysaccharides are usually present beside salts and other compounds, fouling is inevitable. Still, the extent of fouling and its more or less reversible character can be minimized and controlled by using an adequate combination of operating and environmental conditions, and a judicious selection of the membrane to be used. The membrane fouling literature is extensive but a few papers are fundamental for the comprehension of this phenomenon [16–18]. Mitigation of membrane fouling has been addressed through different approaches: (1) optimization of the operating and ﬂuid dynamics conditions; (2) tuning of the environmental conditions of the ﬂuid phase; (3) use of membranes with improved properties. Traditional operation of membrane ﬁltration systems uses pressure differences (typically from 0.5 to 2 bar in ultraﬁltration processes) which promote a signiﬁcant convective transport of media components towards the membrane surface. These transport conditions lead to an increase of solute concentration at the membrane surface and promote solute–solute and solute–membrane interactions that may lead to severe fouling. The understanding of these phenomena supported the concept of critical permeate ﬂux, above which fouling occurs. The critical-ﬂux concept [19–21] explains the success of submerged membrane bioreactors, applied by Kubota and Zenon in wastewater treatment, where the permeate ﬂux is imposed by suction at subcritical or near to critical conditions, assuring low fouling and a long-term operation without the need for membrane cleaning. This improvement is achieved at the expenses of large membrane areas, which became affordable even for wastewater treatment due to the signiﬁcant decrease of membrane cost. Operation under controlled permeate ﬂux, below or near to critical ﬂux conditions, is a strategy that is also used for downstream processing, namely for recovery of proteins from biological media, as will be discussed later in this chapter. It is interesting to note that the problem of operation with large pressure differences between the feed/retentate side and the permeate compartment of membrane ﬁltration modules was identiﬁed long ago. The concept of operation under low uniform transmembrane pressure (UTMP) was pioneered and ﬁrst

j247

j 11 Developments in Membrane Science for Downstream Processing

248

Figure 11.1 Different schemes proposed by Sandblom [22] for permeate recirculation in the shell side of the membrane module in order to assure an approximately constant transmembrane pressure.

patented by Sandblom [22]. UTMP operation makes it possible to beneﬁt from an efﬁcient particle backtransport from the membrane wall at high wall shear rates, while maintaining low TMP in the pressure-dependent regime [23]. Sandblom [22] suggested the use of permeate recirculation in the shell side of the membrane module so that the pressure gradient on the feed/retentate side can be kept approximately constant (Figure 11.1). The TMP can be maintained at a low uniform value throughout the length of the module, independently of the axial velocity in the feed/retentate side. This concept has been revisited recently and considered for protein recovery by membrane processing [23]. Fouling problems have also been tackled through the development of surfacemodiﬁed membranes, namely by covalent attachment of either quaternary amine or sulfonic groups, in order to provide a desired positive or negative charge. These membranes exhibit lower fouling tendency, depending from the environmental conditions tuned for the media to be processed, and may provide high selectivity by exploring both size and charge effects. A recent, and very exciting area of research, deals with the development of membranes that may reverse the hydrophobic/ hydrophilic character of their surface when exposed to an external stimulus. These stimulus-responsive membranes may change their character reversibly when

11.3 Concentration and Purification of Small Bioactive Molecules

exposed to alternate UV/visible light radiation conditions. This approach opens new perspectives for reversing fouling without the use of any external mass agent [24, 25]. 11.2.3 Membrane Selectivity

Selectivity is one of the main issues when assessing the potential use of membranes for downstream processing. Membranes are often regarded as limited in selectivity, when the solutes to be fractionated exhibit close molecular weights and molecular properties. Tuning of the main environmental conditions that characterize the media to be processed may, however, lead to signiﬁcant improvements in selectivity and overall performance. By controlling the pH and ionic strength of the media to be processed it is possible to maximize the differences between the effective volume of the product to be retained and impurities or other products we aim to permeate [26, 27]. As an example, the effective volume of a charged protein accounts for the presence of a diffuse electrical double layer surrounding the protein. Increasing the protein charge, or reducing the solution ionic strength, increases the effective volume thus reducing protein transmission through the membrane. Optimal performance is typically attained by operating close to the isoelectric point of the lower molecular weight protein, to minimize its rejection, and relatively low salt concentrations (10 mM or lower ionic strength) to maximize electrostatic exclusion of the more retained species (e.g., larger protein). Electrical-charge effects can be further exploited by using charged membranes (as referred to above) to increase retention of all species with like polarity. It is important to mention that it may be possible to exploit electrostatic interactions even for solutes with similar isoelectrical points, due to different charge–pH proﬁles for the different species present. The membrane pore-size distribution also affects selectivity by altering the solute sieving coefﬁcients locally. Narrow pore-size distributions, especially for electrically charged membranes, will impact very positively on membrane selectivity and overall performance. In membrane chromatography (see additional discussion below) microﬁltration membranes are modiﬁed by attachment of functional ligands, from the inner pore surface throughout the membrane, in order to provide conditions for highly selective binding interactions with target solutes. In many situations, the reason why membrane chromatography fails to reach commercial application is not related with the intrinsic selectivity of the adsorptive membranes but, rather, due to their dynamic binding capacity that hardly competes with bead-based processes for bind/elute applications [1].

11.3 Concentration and Purification of Small Bioactive Molecules

Recently, the recovery and puriﬁcation of small bioactive molecules from complex media has gained a new interest. Small bioactive molecules comprise a large variety

j249

j 11 Developments in Membrane Science for Downstream Processing

250

Figure 11.2 Diagram relating the size of different particles/ solutes and the corresponding membrane processes.

of compounds with a molecular weight typically below 1 kDa, which includes compounds valuable due to their use as ﬂavors and fragrances, as building blocks or precursors in the ﬁne-chemistry industry, and compounds with antioxidant or anticarcinogenic activity (among other types of desirable biological activity). These compounds are commonly present in complex fermentation media, or even in natural raw materials and subsidiary streams resulting from the processing of these materials. Their recovery is usually difﬁcult due to their low concentration, often vestigiary, and the complexity of the original matrix where they have to be recovered from. This chapter discusses, and illustrates with recent applications, the use of different membrane processes able to deal with the recovery of small biologically active molecules (see Figure 11.2): electrodialysis, pervaporation, and nanoﬁltration. 11.3.1 Electrodialysis

The most interesting examples of the use of electrodialysis for recovery of target small molecules are related with the in-situ integration of this technique during biotransformations. It is often observed that the ﬁnal product of a biotransformation process causes inhibition of the biocatalyst involved, even at relatively low concentrations (<500 mmol L1). To overcome this problem it is necessary to recover the product formed in an integrated process, in order to keep its concentration within a desirable concentration level. Jonsson and coworkers developed and patented a process [28] which allows for the recovery of lactic acid from an active fermentation by integrating in situ an electroenhanced dialysis step followed by bipolar electrodialysis where the ionic species are concentrated. This procedure allows for a continuous production and recovery of lactate, while simultaneously controlling the fermentation pH at its optimal level. Another interesting integrated approach has

11.3 Concentration and Purification of Small Bioactive Molecules

been described by Zelic et al. [29], where an electrodialysis step is used for the continuous recovery of pyruvate from an active fermentation process. The simultaneous separation and recovery of acidic and basic bioactive peptides by employing electrodialysis with ultraﬁltration membranes has also been investigated recently [30]. This work aims at demonstrate the feasibility of separating peptides from a beta-lactoglobulin hydrolysate, using an ultraﬁltration membrane stacked in an electrodialysis cell, and a study of the effect of pH on the migration of basic/ cationic and acid/anionic peptides in the electrodialysis conﬁguration. 11.3.2 Pervaporation

The recovery of ﬂavors and fragrances from diluted aqueous streams may be of industrial interest under different circumstances: (1) recovery of complex aroma proﬁles and/or target aroma compounds from active biocatalytic processes; (2) recovery of complex aroma proﬁles and/or target aroma compounds from natural extracts and industrial processes aqueous streams. Pervaporation offers a unique solution for the recovery of complex aroma proﬁles. An example for the recovery of complex aroma proﬁles, faithful to their origin, is the recovery of a muscatel aroma from an ongoing wine-must fermentation [31, 32]. Coupling pervaporation to active bioconversion processes is extremely interesting because it allows for continuous removal of target compounds that, otherwise, may simultaneously exert an inhibitory effect over the biocatalyst. Several examples have been discussed in the literature [33, 34] referring to the advantages of integrating bioconversion processes and pervaporation. However, not much has been discussed about the problem of production of noncondensable gases during biological processes (namely carbon dioxide), which permeate the membrane. The presence of noncondensable gases requires an additional energy input in order to keep the downstream pressure at desirable levels and leads to a decreasing energy efﬁciency of the condensation process. Although membranes may exhibit a high afﬁnity towards aroma compounds, the high diffusivity of water, even through hydrophobic membranes, limits the degree of selectivity for aroma recovery from diluted aqueous media, such as fermentation media and most natural matrices and process streams. The membrane material of choice for organophilic pervaporation has been polydimethylsiloxane (PDMS), including chemically modiﬁed derivates with bulky side groups, introduced in order to reduce water ﬂux (such as polyoctylmethylsiloxane – POMS) and other elastomeric materials such as polyether-polyamide block-copolymers (PEBA), ethylene-propylene-diene monomer (EPDM) elastomers and ﬁller-type membranes [35]. Because one aims at employing selective membranes that are as thin as possible, in order to have a high sorption afﬁnity and minimize the relevance of diffusion selectivity, most organophilic pervaporation membranes are composites consisting of a thin selective active layer and a macroporous support, which assures mechanical stability. Mass-transfer limitations due to poor hydrodynamic conditions in the feed-side/ membrane interface are common in organophilic pervaporation (as referred above).

j251

j 11 Developments in Membrane Science for Downstream Processing

252

Figure 11.3 Partial fluxes of isoamyl alcohol, ethyl acetate, isoamyl acetate and ethyl hexanoate as a function of their feed crossflow velocity (bottom axis) and Reynolds number (top axis) in a singlechannel module, using a POMS-PEI composite membrane. Notice that external mass-transfer limitations are not fully overcome when solutes with a high affinity towards the membrane are processed (Adapted from Ref. 32.)

This effect, usually known as feed-side concentration polarization, may become particularly relevant for solutes with a high sorption afﬁnity towards the membrane, which may lead to its depletion near the membrane interface if external mass-transfer conditions are not sufﬁciently good to guarantee their fast transport from the bulk feed to the interface [32, 36] (see Figure 11.3). As a consequence of their depletion near the interface the driving force for transport, and the resulting partial ﬂuxes, become lower. Product recovery and capture can be carried out in a series of condensation stages, at different temperatures, in order to achieve different fractions enriched in target compounds. The temperature of each condenser has to be adjusted according to the downstream pressure in the circuit and the character of the compounds to be separated and recovered [37, 38]. Capture of the target-permeating compounds by condensation remains one of the main problems for competitive use of pervaporation systems, due to the energy costs involved to keep an adequate downstream pressure and to cool down the permeating stream.

11.3 Concentration and Purification of Small Bioactive Molecules

New ways of capturing the permeating vapors have to be developed, in order to render this process competitive, enabling continuous operation and reducing energy input. One of the most interesting approaches under development involves the capture of the target aroma compounds by promoting their incorporation (i.e. by solubilization) into a designed system, which will be used as a delivery agent directly in the ﬁnal food, cosmetic or pharmaceutical product. This is a very effective and elegant way to use the same system to, ﬁrstly, capture the aromas from the permeating vapor stream and, secondly, to deliver them into the end product. Most research on aroma recovery by organophilic pervaporation has been conducted using aqueous aroma model solutions [39, 40], although in recent years a signiﬁcant interest has been devoted to the recovery of aroma compounds from natural complex streams, such as fruit juices [41, 42], subsidiary streams from the food industry [43] and other natural matrices [44]. The increasing demand for natural aroma compounds for food use, and their market value, opens a world of possibilities for a technique that allows for a benign recovery of these compounds without addition of any chemicals or temperature increase. Considering the strong growth predicted for the low-fat and low-sugar foods and beverages market, the global demand for ﬂavors is expected to grow signiﬁcantly. Hence, the development of new technologies and delivery systems that improve the use of ﬂavors in food products is likely to be crucial to the future development of this market. Organophilic pervaporation may play an important role for replacement of evaporative techniques as well as aroma recovery processes based on solvent extraction, in particular when the label natural is considered crucial. The technical challenges discussed above have to be addressed in order to render organophilic pervaporation a competitive process, in particular, the way of capturing target aromas from the permeate stream has to be reassessed in terms of energy consumption and labor intensity. 11.3.3 Nanofiltration

Nanoﬁltration is a pressure-driven membrane process that lies between ultraﬁltration and reverse osmosis (see Figure 11.2) in terms of its ability to reject molecular or ionic species. Usually it is considered that nanoﬁltration membranes may exhibit nominal cutoffs between 1000 and 200 Da, but this classiﬁcation should not be regarded as a clear and sharp domain between ultraﬁltration (larger cutoffs) and reverse osmosis (smaller cutoffs). The membranes used in nanoﬁltration – most are polymeric membranes because ceramic nanoﬁltration membranes are still in their infancy – may be so dense that pores cannot be regarded as such and tighter nanoﬁltration membranes have to be considered as dense, with the molecular transport taking place through free-volume elements of the polymer. Nanoﬁltration membranes usually provide for a good retention of small organic molecules and inorganic salts, especially if multivalent ions are involved. The high rejection of these compounds may lead to signiﬁcant osmotic pressure differences across the membrane, which decreases the actual driving force for

j253

j 11 Developments in Membrane Science for Downstream Processing

254

transport. Therefore, in order to assure high solvent ﬂuxes and good solute(s) rejection(s), these membranes should exhibit a high afﬁnity for the solvent (hydrophilic membranes for processing of aqueous streams) and low afﬁnity for the solute(s). Taking into consideration its rejection characteristics nanoﬁltration has been proposed for the recovery and fractionation of bioactive molecules with molecular weight lower than 1 kDa from complex media. Recent work describes the successful use of nanoﬁltration for the recovery of biologically active oligosaccharides from milk [45], using a combination of enzymatic treatment of defatted milk with betagalactosidase and nanoﬁltration. It was shown that enzymatic hydrolysis of lactose signiﬁcantly improves the efﬁciency and selectivity of this process. The human milk oligosaccharides recovered by this method were shown to inhibit binding of intimin, an adhesion molecule of enteropathogenic Escherichia coli, to epithelial cells in vitro. No signiﬁcant difference was found in the oligosaccharide proﬁle between samples prepared by this method and conventional gel-permeation chromatography. The approach developed was also shown to be suitable for the recovery of substantial quantities of tri- and tetra-saccharides from caprine milk. A similar concept has been proposed and applied to the recovery of bioactive peptides, namely peptides derived from different proteins from bovine colostrum, milk, and cheese whey [46], for human nutrition and promotion of human health. Active peptides can be liberated during gastrointestinal digestion or milk fermentation with proteolytic enzymes. Such peptides may exert a number of physiological effects in vivo on the gastrointestinal, cardiovascular, endocrine, immune, nervous, and other body systems. A number of bioactive peptides have been identiﬁed in fermented dairy products, and there are already a few commercial dairy products enriched with blood-pressure-reducing milk protein peptides. However, the industrial-scale production of such peptides has been limited by a lack of suitable technologies. Nanoﬁltration offers the potential for recovering of these peptides and fractionating them according to their molecular weight and charge, excluding higher molecular weight fractions. Size exclusion is not the sole mechanism to be explored in order to achieve a desirable selectivity: membrane–solute electrostatic effects may be explored favorably by playing with the selection of the membrane material and the environmental conditions of the media to be processed. Recovery of valuable bioactive compounds by nanoﬁltration, from natural products or streams resulting from the processing of natural products, is also gaining an increasing interest. Recent examples include the production of natural extracts from olive oil subproducts, which are rich in the most potent natural antioxidant compound identiﬁed so far (hydroxytyrosol) as well as the production of natural extracts from grape pomace residues, which are rich in a number of high-value compounds. The ﬁrst recognized properties of hydroxytyrosol were its ability to prevent the oxidation of the low-density lipoprotein (LDL) [47] and the aggregation of blood platelets [48]. Manna et al. [49] proved that this compound is able to protect several cellular human systems from the toxicity induced by reactive oxygen species.

11.4 Concentration and Purification of Large Bioactive Molecules

Figure 11.4 Process diagram for the production of hydroxytyrosolrich natural extracts produced from olive-oil solid residues.

Hydroxytyrosol has been also referred to as a potent agent with anticarcinogenic activity. Grape pomace is rich in a large number of polyphenolic compounds with high antioxidant activity, with resveratrol being probably the most known and referred to in the scientiﬁc literature. In both cases, solid residues are extracted either with water or hydro-alcoholic solutions at mild temperatures (an enzymatic pretreatment may also be added in order to break glycosidic bonds), and the supernatant is processed by nanoﬁltration (Figure 11.4). This operation allows for recovering a fraction of compounds within a desirable range of molecular weights, while rejecting higher molecular weight compounds (frequently with an undesirable biological activity) such as pesticides and heavy metals, very efﬁciently excluded by nanoﬁltration [50]. This approach can be extended to a large number of natural products rich in compounds with desirable biological activity, which include fruits and other plants – for example, the recovery of phytosterols and tocopherol from vegetable oils, as well as natural marine products.

11.4 Concentration and Purification of Large Bioactive Molecules

As noted above, most review papers dealing with downstream processing are focused on the recovery and puriﬁcation of large bioactive molecules, namely proteins. Early biotechnology products were highly active hormones (e.g., insulin, human growth hormone, erythropoietin), thrombolytic agents (e.g., tissue-type plasminogen activator) and clotting factors (e.g., factor VIII). These compounds were typically produced on quite a small scale, considering the annual production requirements ranging from 1 to 10 kg. Recent products are monoclonal antibodies, used for the treatment of breast cancer, B-cell lymphoma and rheumatoid arthritis, among others.

j255

j 11 Developments in Membrane Science for Downstream Processing

256

These molecules act stoichiometrically, binding to a particular receptor or cell type, thus requiring much higher dosing levels and batch sizes, in order to satisfy annual production requirements of the order of 1000 kg [1]. As a consequence, process optimization and, in particular, downstream processing is become more relevant in terms of productivity and costs of production. The excellent review by van Reis and Zydney [1] provides a comprehensive discussion of all major uses of membranes for processing of large molecules. This chapter will be essentially focused on the use of ultraﬁltration for the concentration and fractionation of proteins, and the development of membrane chromatography systems. 11.4.1 Ultrafiltration

Ultraﬁltration is mostly used for protein concentration and buffer exchange at the industrial scale, replacing size-exclusion chromatography. For protein concentration the selection of membranes with low nominal cutoff values is usually recommended, although, as previously discussed, high rejection and controlled fouling operation may be achieved by using charged membranes together with a correct adjustment of the environmental conditions (pH and ionic strength) of the media to be processed. Recently, the term high-performance tangential ﬂow ﬁltration (HPTFF) became common to describe the operation of ultraﬁltration processes under optimized conditions, with a particular emphasis on protein fractionation [51]. Recent results obtained with HPTFF [1] have shown the potential of this process for the fractionation of monomers from oligomers [51], protein variants differing at only a single amino acid residue [52], and an antigen binding fragment from a similar size impurity [53]. The improved selectivity achieved in HPTFF can only be obtained through the understanding of the phenomena taking place when the solute(s) to be processed interacts with the membrane surface (Figure 11.5). A series of optimized

Figure 11.5 Scheme representing the different forces acting on a solute near a membrane surface, during a convective filtration process.

11.4 Concentration and Purification of Large Bioactive Molecules

approaches, developed in recent years and partially discussed above in this chapter, can therefore be implemented: (1) operation in the pressure-dependent regime, below a critical ﬂux, employing the concept of low uniform transmembrane pressure (UTMP); (2) adjustment of the environmental conditions of the media to be processed by adequate regulation of its pH and ionic strength; (3) use of electrically charged membranes, if possible with a narrow pore-size distribution, in order to increase the retention of species with like polarity. Overall process optimization is achieved by determining the best compromise between yield and purity, as a function of selectivity and mass throughput [54]. As previously discussed, operation under controlled permeate ﬂux (in opposition to operation under controlled transmembrane pressure) offers a high degree of process control, making it possible to operate under gentle convective transport conditions, which minimize solute transport to the membrane surface keeping osmotic pressure differences low and reducing fouling. Figure 11.6a and b [55]

Figure 11.6 Fractionation of b-lactoglobulin and g-globulin (detection by fluorescence) through a prefouled 50-kDa PES membrane under controlled permeate flux operation (a). Fractionation of b-lactoglobulin and g-globulin through a prefouled 50-kDa PES membrane under controlled transmembrane pressure operation (b). Adapted from Ref. 55.

j257

j 11 Developments in Membrane Science for Downstream Processing

258

compare the fractionation of gamma-globulin (155 kDa) and beta-lactoglobulin (monomer with 18.3 kDa, although it forms dimers and tetramers easily, according to the environmental media conditions) using a polyethersulfone membrane with a nominal cutoff of 50 kDa. As can be seen, for a controlled permeate ﬂux of 25 l m2 h1 and above this critical value, fouling occurs and the resulting transmembrane pressure increases (Figure 11.6a). Still, rejection of gamma-globulin is as high as 98.2%, while rejection of beta-lactoglobulin is always lower than 63%; operation under these conditions allows for a good fractionation of the two proteins if a cascade procedure is employed, because a good rejection of the larger protein is achieved. In contrast, if a controlled transmembrane-pressure strategy is used (Figure 11.6b), even under low TMP conditions (between 0.15 and 0.25 bar), rejection of the larger protein decreases signiﬁcantly up to values between 75 and 70% without a signiﬁcant improvement in the transmission of beta-lactoglobulin. As a consequence, fractionation of the two proteins becomes quite ineffective. Additionally, under controlled-pressure operation, fouling became quite signiﬁcant and the permeate ﬂux declined dramatically with time. Operation under controlled TMP with low pressure differences, comparable with the ones achieved during controlled ﬂux operation, could be a strategy to follow but in traditional ultraﬁltration modules it is not possible to assure such a uniform and low TMP. Since selectivity is a function of the local permeate ﬂux, and thus the local transmembrane pressure, selectivity can be further improved by maintaining a nearly uniform and low transmembrane pressure throughout the ultraﬁltration module. Following the early work from Sandblom, HPTFF technology uses, when required, a cocurrent permeate ﬂow that is accomplished by the addition of a coﬂow loop and pump on the permeate side. During permeation, proteins are exposed to different processing conditions, such as shear stress (e.g., when permeating the membrane pores) and interaction with the membrane and the membrane pores surface, which may induce reversible or irreversible changes in the protein structure. In fact, permeation of protein molecules through membranes with a molecular weight cutoff lower than their mass has been reported (see Figure 11.6 showing permeation of gamma-globulin through a 50-kDa membrane). This permeation may be attributable to pore-size distribution (e.g., the presence of pores with a higher diameter than the nominal cutoff), or to the protein shaping and orientation, which may favor its passage through the membrane pores. However, since distinct conformational states of proteins possess small energetic differences, the hypothesis of protein structural alteration (e.g., molecular elongation) during the permeation process, which facilitates their passage through the membrane pores, has to be considered. The impact of parameters such as the ratio of protein mass to the membrane cutoff (l), or the effect of the membrane material during permeation, on their structure/ function, is currently unknown and constitutes an essential requisite for the full development and implementation of membrane processes for the selective fractionation of proteins. All these alterations, which may be induced when a protein molecule interacts with a membrane surface either during convective (ﬁltration) processes or under diffusive

11.4 Concentration and Purification of Large Bioactive Molecules

Figure 11.7 Explanatory diagram showing the use of Tryptophan as an intrinsic fluorescence probe for proteins.

conditions (diaﬁltration), can be reported by intrinsic ﬂuorescence probes such as tryptophan residues and detected by using appropriate natural ﬂuorescence techniques (see explanatory scheme in Figure 11.7). One of the most interesting features of natural ﬂuorescence results from the fact that the ﬂuorescence response of a given molecule depends very much on their microenvironment. This feature can be used in order to gather information about the structure of complex molecules such as polypeptides and proteins, which may integrate several ﬂuorescent amino acids residues such as tryptophan, tyrosine, and phenylalanine. Among these, tryptophan is the one that exhibits the highest quantum yield, which makes it a good candidate to be used as an intrinsic ﬂuorescence reporter. The basic concept is the use of the ﬂuorescence response of tryptophan residues, embedded in the polypeptide/protein structure, which is sensitive to changes in its microenvironment. Therefore, if a given protein interacts with a membrane surface and, due to this interaction, changes its tridimensional structure, it can be anticipated that the relative position of the tryptophan residue(s) may be altered. Such changes, may include: (1) a higher exposure of a buried tryptophan to the surrounding solvent (usually water) due to unfolding processes, which leads to contact with a more hydrophilic environment; (2) the opposite process, involving the movement of a tryptophan residue to a more buried, hydrophobic environment; (3) the increase of a tryptophan residue mobility, which will occur if the protein assumes a more unfolded conformation in the region where the tryptophan is located; (4) the change of the relative position of a tryptophan residue towards internal protein quenchers, such as aspartate residues and disulﬁde bonds, which may occur as a result of processes of folding/unfolding.

j259

j 11 Developments in Membrane Science for Downstream Processing

260

In a recent series of papers [56–59], using complementary information acquired by different natural ﬂuorescence techniques, such as steady-state ﬂuorometry, steady-state ﬂuorescence anisotropy and time-decay ﬂuorescence, it was shown that the ratio of protein mass to the membrane cutoff (l) may determine signiﬁcant changes on the protein conformation leading to its unfolding under higher pore mouth and intrapore stress conditions. It was also shown that selection of the membrane material can be a key issue, because some materials may exhibit a high afﬁnity for the metal center of enzymes, such as the C isoenzyme of horseradish peroxidase, where the prosthetic heme group partially loses its protoporphyrin ring coordinating iron ion, leading to a deep change of its structure and, ultimately, loss of activity. The use of these natural ﬂuorescence techniques offers not only the possibility of studying the interaction of proteins with membranes, under convective and diffusive conditions, but also they may be easily extended to studies involving proteins and other porous materials such as chromatography media. The areas of application of these techniques will range from polypeptide and protein fractionation to the monitoring of systems where protein–surface interactions are relevant. 11.4.2 Membrane Chromatography

A large number of membrane materials containing functional ligands has been developed, namely ion-exchange, afﬁnity, reversed-phase, and hydrophobic interaction membranes. Although the binding capacity in membranes tends to be low, the convective ﬂow through the pores reduces mass-transfer resistance compared with bead chromatography. The main beneﬁt of membrane chromatography is associated with the shorter transport times achieved, in comparison with bead chromatography, as the interaction between molecules and active sites at the membrane surface occurs in a convective through pore ﬂow, rather than in a stagnant ﬂuid phase inside the pores of an adsorbent particle. This feature can be particularly important for puriﬁcation of large biomolecules and viruses that may exhibit signiﬁcant transport limitations in conventional chromatographic media. Membrane chromatography also offers higher ﬂow rates, lower pressure drops, and shorter processing times than traditional chromatography. Their use is also accompanied by reduced protein unfolding and denaturation, as (discussed above) [3]. Taking these features together, membrane chromatography is emerging as an alternative for ﬂow-through applications, where a rapid processing of high volumetric and dilute feed streams is required [60]. Flow-through applications include also the removal of DNA, viruses, and endotoxins [1]. In addition, it has been shown that membrane adsorbers are competitive in bindand-elute applications for large solutes such as DNA, RNA, and viruses. Most bead chromatographic media have pore sizes that are too small, which exclude large molecules from entering and binding to speciﬁc sites. Under these circumstances, membrane chromatography exhibits a competitive binding capacity for these molecules, such as DNA.

11.5 Future Trends and Challenges

11.5 Future Trends and Challenges

The constraints discussed in this chapter represent new challenges and opportunities for improved membrane materials and technologies for downstream processing. The use of membrane processes for the recovery of small biologically active molecules is expected to grow signiﬁcantly within the next years, in particular in what refers to the use of nanoﬁltration for recovery of high added value compounds with impact on human nutrition and promotion of human health. Some of the recovery processes will demand membranes that are more stable in hydro-alcoholic media, or even in organic media compatible with the ﬁnal product and its use, exhibiting simultaneously a sharp cutoff behavior. The discovery of small molecules with desirable biological properties, present in natural matrices such as plants and marine products, will boost the need for recovery processes regarded as clean and sustainable that allow the use of the label natural in the ﬁnal product. Membrane processes ﬁt perfectly this demand due to the mild conditions under which they operate. In what concerns ultraﬁltration, it has replaced size-exclusion chromatography in almost all ﬁnal formulation processes. Charged ultraﬁltraﬁon membranes, in conjunction with optimum operating parameters, as previously discussed, can also be used to enable protein puriﬁcation with HPTFF. In fact, recent developments in membrane chromatography and HPTFF enable, for the ﬁrst time, complete puriﬁcation of proteins using membrane systems [1]. Membrane chromatography has been evolving slowly and its low binding capacity has been the major obstacle. To be competitive for applications in which the product is bounded and then eluted, membranes will need to have a binding capacity equivalent to that of bead chromatography, and similar process time and cost. On the other hand, ﬂow-through chromatography applications that involve the binding of impurities, whereas the products ﬂow through the matrix, are nowadays competitive. One interesting issue is the development of disposable systems (and disposable, single-use membranes). Disposable systems may become attractive for production processes, because they eliminate the need for development and validation of cleaning cycles. In this case, the development of biodegradable membranes will become an interesting opportunity. In many cases, the improvement in external mass transfer is achieved at the expense of membrane packing density. This is the case for a number of solutions involving the use of spacers or, for example, the case of rotating modules that promote the formation of Taylor vortices. As mentioned above, membranes with micro- or nanostructured 3D topography offer an extremely elegant solution for improvement of mass transfer by inducing turbulence at the membrane scale, avoiding the spending of energy in the bulk ﬂuid and allowing for the construction of highly packed modules. In fact, fascinating breakthroughs are expected in the coming years, through the development of membranes designed to operate in a particular environment,

j261

j 11 Developments in Membrane Science for Downstream Processing

262

respond to speciﬁc controlled stimulus (pH, ionic strength, temperature, UV/Vis radiation or magnetic ﬁeld) or interact speciﬁcally through molecular recognition with target solutes.

References 1 van Reis, R. and Zydney, A. (2007) Journal of Membrane Science, 297, 16–50. 2 Hubbuch, J. and Kula, M.-R. (2007) Journal of Non-Equilibrium Thermodynamics, 32, 99–127. 3 Charcosset, C. (2006) Biotechnology Advances, 24, 482–492. 4 Giorno, L. and Drioli, E. (2000) Trends in Biotechnology, 18, 339–348. 5 Lopez, J.L. and Matson, S. (1997) Journal of Membrane Science, 125, 189–211. 6 Sousa, H.A., Afonso, C.A.M., Mota, J.P.B. and Crespo, J.G. (2005) Chemical Engineering Research and Design, 83 (A3), 285–294. 7 Vogelaar, L., Lammertink, R.G.H. and Wessling, M. (2006) Langmuir, 22, 3125–3130. 8 Susanto, H. and Ulbricht, M. (2007) Langmuir, 23, 7818–7830. 9 He, D.M. and Ulbricht, M. (2007) Macromolecular Chemical Physics, 208, 1582–1591. 10 Gronda, A.M., Buechel, S. and Cussler, E.L. (2000) Journal of Membrane Science, 165 (2), 177–187. 11 Belfort, G., Pimbley, J.M., Greiner, A. and Chung, K.-Y. (1993) Journal of Membrane Science, 77, 1–22. 12 Kaur, J. and Agarwal, G.P. (2002) Journal of Membrane Science, 196, 1–11. 13 Kelder, J.D.H., Janssen, J.J.M. and Boom, R.M. (2007) Journal of Membrane Science, 304 (1–2), 50–59. 14 Santos, J.L.C., Geraldes, V., Velizarov, S. and Crespo, J.G. (2007) Journal of Membrane Science, 305 (1–2), 103–117. 15 Kane, R.S., Takayama, S., Ostuni, E., Ingber, D.E. and Whitesides, G.M. (1999) Biomaterials, 20, 2363–2376.

16 Bacchin, P., Aimar, P. and Sanchez, V. (1995) AIChE Journal, 41, 368–377. 17 Belfort, G., Davis, R.H. and Zydney, A.L. (1994) Journal of Membrane Science, 96, 1–58. 18 Mcdonogh, R.M., Fane, A.G. and Fell, C.J.D. (1989) Journal of Membrane Science, 43, 69–85. 19 Field, R.W., Wu, D., Howell, J.A. and Gupta, B.B. (1995) Journal of Membrane Science, 100, 259–272. 20 Howell, J.A. (1995) Journal of Membrane Science, 107, 165–171. 21 Bacchin, P., Aimar, P. and Field, R.W. (2006) Journal of Membrane Science, 281 (1–2), 42–69. 22 Sandblom, R.M. (1978) SW patent 74,16,257, 1974 and US patent 4,105,547. 23 Baruah, G.L., Nayak, A. and Belfort, G. (2006) Journal of Membrane Science, 274, 56–63. 24 Lim, H.S., Han, J.T., Kwak, D., Jin, M. and Cho, K. (2006) Journal American Chemical Society, 128, 14458–14459. 25 Nayak, A., Liu, H. and Belfort, G. (2006) Angewandt Chemie International, 45, 4094–4098. 26 Pujar, N.S. and Zydney, A.L. (1998) Journal of Chromatography A, 796, 229–238. 27 Saksena, S. and Zydney, A.L. (1994) Biotechnology & Bioengineering, 43, 960–968. 28 Garde, A., Rype, J.-U. and Jonsson, G. (2002) WO patent 02/48044 A2. 29 Zelic, B., Gostovic, S., Vuorilehto, K., Vasic-Racki, D. and Takors F R. (2004) Biotechnology & Bioengineering, 85 (6), 638–646.

References 30 Poulin, J.F., Amiot, J. and Bazinet, L. (2006) Journal of Biotechnology, 123 (3), 314–328. 31 Sch€afer, T., Bengtson, G., Pingel, H., B€ oddeker, K.W. and Crespo, J.G. (1999) Biotechnology & Bioengineering, 62 (4), 412–421. 32 Sch€afer, T. and Crespo, J.G. (2007) Journal of Membrane Science, 301 (1–2), 46–56. 33 Stefer, B. and Kunz, B. (2002) Chemie Ingenieur Technik, 74 (7), 1029–1034. 34 Maume, K.A. and Cheetham, P.S.J. (1991) Biocatalysis, 5, 79. 35 Rutherford, S.W., Kurtz, R.E., Smith, M.G., Honnell, K.G. and Coons, J.E. (2005) Journal of Membrane Science, 263 (1–2), 57–65. 36 Sch€afer, T., Vital, J. and Crespo, J.G. (2004) Journal of Membrane Science, 241 (2), 197–205. 37 Marin, M., Hammami, C. and Beaumelle, D. (1996) Journal of Food Engineering, 28 (3–4), 225–238. 38 Br€ uschke, H.E.A., Schneider, W. and Tusel, G.F. (1989) EP patent 0 332 738. 39 Baudot, A. and Marin, M. (1997) Food and Bioproducts Processing, 75 (C2), 117–142. 40 Borjesson, J., Karlsson, H.O.E. and Tragardh, G. (1996) Journal of Membrane Science, 119 (2), 229–239. 41 Pereira, C.C., Ruﬁno, J.R.M., Habert, A.C., Nobrega, R., Cabral, L.M.C. and Borges, C.P. (2005) Journal of Food Engineering, 66 (1), 77–87. 42 Alvarez, S., Riera, F.A., Alvarez, R., Coca, J., Cuperus, F.P., Bouwer, S.T., Boswinkel, C., van Gemert, R.W., Veldsink, J.W., Giorno, L., Donato, L., Todisco, S., Drioli, E., Olsson, J., Tragardh, G., Gaeta, S.N. and Panyor, L. (2000) Journal of Food Engineering, 46 (2), 109–125. 43 Souchon, I., Pierre, F.X., Athes-Dutour, V. and Marin, A. (2002) Desalination, 148 (1–3), 79–85. 44 Kattenberg, H.R. and Willemsen, J.H.A. (1999) WO patent 00/38540. 45 Sarney, D.B., Hale, C., Frankel, G. and Vulfson, E.N. (2000) Biotechnology and Bioengineering, 69 (4), 461–467.

46 Butylina, S., Luque, S. and Nystrom, M. (2006) Journal of Membrane Science, 280 (1–2), 418–426. 47 Visioli, F. and Galli, C. (1998) Journal of Agricultural and Food Chemistry, 46 (10), 4292–4296. 48 Petroni, A., Blasevich, M., Salami, M., Papini, N., Montedoro, G.F. and Galli, C. (1995) Thrombosis Research, 78 (2), 151–160. 49 Manna, C., Galleti, P., Maisto, G., Cucciola, V., DAngelo, S. and Zappia, V. (2000) FEBS Letters, 470 (3), 341–344. 50 Da Ponte, M.N., Santos, J.L., Matias, A., Nunes, A., Duarte, C. and Crespo, J.G. (2007) patent WO2007013032. 51 van Reis, R., Gadam, S., Frautschy, L.N., Orlando, S., Goodrich, E.M., Saksena, S., Kuriyel, R., Simpson, C.M., Pearl, S. and Zydney, A.L. (1997) Biotechnology and Bioengineering, 56 (1), 71–82. 52 Ebersold, M.F. and Zydney, A.L. (2004) Biotechnology Progress, 20 (2), 543–549. 53 van Reis, R., Brake, J.M., Charkoudian, J., Burns, D.B. and Zydney, A.L. (1999) Journal of Membrane Science, 159 (1–2), 133–142. 54 van Reis, R. and Saksena, S. (1997) Journal of Membrane Science, 129 (1), 19–29. 55 Crespo, J.G., Trotin, M., Hough, D. and Howell, J.A. (1999) Journal of Membrane Science, 155 (2), 209–230. 56 Portugal, C.A.M., Crespo, J.G. and Lima, J.C. (2006) Journal of Photochemistry and Photobiology B-Biology, 82 (2), 117–126. 57 Portugal, C.A.M., Lima, J.C. and Crespo, J.G. (2006) Journal of Membrane Science, 284 (1–2), 180–192. 58 Portugal, C.A.M., Crespo, J.G. and Lima, J.C. (2007) Journal of Membrane Science, 300 (1–2), 211–223. 59 Portugal, C.A.M., Lima, J.C. and Crespo, J.G. (2008) Journal of Membrane Science, 321 (1), 69–80. 60 Gottschalk, U. (2005) Biopharm International, 18 (3), 42–58.

j263

j265

12 Integrated Membrane Processes Enrico Drioli and Enrica Fontananova

12.1 Introduction

Separation processes are intensively used in practically all industrial processes. About 40–50% of the energy use in major commodity-producing industries is utilized in separations, but most of these are still carried out by traditional thermally driven processes [1]. The necessity to realize a sustainable growth, also by a more rational and efﬁcient energy use, calls for additional and substantial developments in the separations ﬁeld. Sustainable growth focuses on making progress to satisfy global human needs without damaging the environment. A promising way for the realization of a sustainable growth, is the strategy of process intensiﬁcation [2]. This consists of innovative equipments design and process development methods that are expected to bring relevant improvements in manufacturing and processing, decreasing production costs, energy consumption, waste generation, equipment size and improving remote control, scale-up, design and process ﬂexibility resulting in cheaper, sustainable technical solutions [2]. Because of its intrinsic properties that well ﬁt the requirements of processintensiﬁcation strategy (efﬁciency, modularity, reduced energy consumption, etc.), membrane-separation processes have well-established applications in various industrial ﬁelds and more progresses can be anticipated for the near future [3]. Various membrane operations are available today for a wide spectrum of industrial applications. Microﬁltration (MF), ultraﬁltration (UF), nanoﬁltration (NF), reverse osmosis (RO), gas and vapor separation (GS, VS), pervaporation (PV), dialysis (D), electrodialysis (ED) and membrane contactors (MCs) are only some of the bestknown membrane unit operations. However, the possibility to integrate various membrane operations in the same process or in combination with conventional separation units, allows, in many cases better performance in terms of product quality, plant compactness, environmental impact, and energy use to be obtained.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 12 Integrated Membrane Processes

266

In this chapter, some important examples of integrated membrane systems will be presented starting from the most well-known application of membrane technology, that is, water desalination.

12.2 Integrated Membrane Processes for Water Desalination

Membrane technology is already recognized as the best choice for water desalination since thermal options for desalting are about 10 times less energy efﬁcient [1, 4]. The number of membrane desalination installations accounts for about the 80% of the total number of desalination plants [5] and for about the 50% of the total capacity [5] (Table 12.1). Although RO water desalination is today considered as the most cost-effective solution [4], key factors for further improvements in membrane-based desalination systems are: enhancement of water-recovery factor, cost reduction, improvement of water quality, new brine-disposal strategies. All these issues can be addressed by an integrated approach. Traditional pretreatments make an extensive use of chemicals (NaClO as disinfection, FeCl3 as ﬂocculants, H2SO4 as antiscaling agent) and mechanical ﬁltration units (sand ﬁltration, media ﬁltration, cartridge ﬁltration). Another interesting possibility is the use of pressure-driven membrane processes, in particular MF and UF are becoming standard and very efﬁcient pretreatment options for sea- and brackish-water desalination. Also, for wastewater treatment, MF/UF pretreatment technology can efﬁciently reduce the highly fouling nature of the feed. UF is typically used to retain macromolecules, colloids, solutes with molecular weight higher than a few thousand Daltons. MF is a low-pressure membrane process for separating colloidal and suspended micrometer-size particles [6]. Table 12.1 Membrane technology vs. thermal technologies for water desalination (Data from Ref. 5.)

Desalination plant type Thermal desalination

Membrane desalination

Multistage ﬂash (MSF) Vapor compression (VC) Multieffect distillation (MED) Reverse osmosis (RO) Electrodialysis reversal (EDR)

% over the total number of desalination plants worldwide (14 000)

% over the total capacity of desalination plants worldwide (7 000 000 MGD)

20

50

80

45 30 25 90 10

50

85 10 5 90 10

12.2 Integrated Membrane Processes for Water Desalination

Although the capital cost of membrane pretreatment is usually higher than that of a conventional pretreatment, the additional cost of MF/UF is paid and also exceeded by reducing RO replacement and chemical cost for both dosing and RO cleaning [7]. Other potential beneﬁts arise from the 33% space saving of MF/UF and the opportunity to increase RO ﬂux and water recovery. Moreover MF/UF provides a more reliable system, and is tolerant of feed-quality variations [7]. An UF system utilizing hollow-ﬁber (HF) membranes has been successfully used as pretreatment prior to seawater reverse osmosis (SWRO) desalination without any chemical treatments [8]. The quality of UF permeate was good and satisﬁed the need of SWRO feed water [8]. In this pilot plant the UF pretreatment system is arranged in 2 trains, each housing 3 modules (PAN HF membranes, nominal pore size of 0.02 mm, MWCO 50 000, total effective surface area of 30 m2). Raw seawater (samples from Qingdao Jiaozhou Bay, the Yellow Sea of China) was ﬁrst passed into a cartridge sand ﬁlter and successively feed to UF system, the UF permeate was then pumped to the RO system (spiralwound composite polyamide) (Figure 12.1). The advantages of UF pretreatment in comparison with traditional ones, has been also examined for an Eastern Mediterranean feed water [7]. In this study it has been shown that the UF/MF can be cheaper than conventional pretreatment (dual-media ﬁlters followed by cartridges) by 0.7 cents/m3 for total water cost [7]. Although the integration of RO with other pressure-driven membrane processes has led to signiﬁcant improvements in membrane-based desalination process economics, another fundamental problem is the environmental aspects of brine discharge from reverse-osmosis desalination plants. The most frequent disposal practice is the direct discharge in the sea. However, the more and more stringent environmental regulations preclude, in many cases, this low-cost possibility in order to protect the aquatic environment. Various process engineering strategies have been investigated in order to have a more environmentally friendly strategy for brine disposal in reverse-osmosis desalination. A suitable solution is the possibility to redesign completely a desalination system by also introducing MC operations [9–11]. Membrane contactors are systems in which the membrane function is to facilitate diffusive mass transfer between two contacting phases (liquid–liquid, liquid–gas, etc.) without dispersion of one phase within another [12]. The membrane does not act as a selective barrier, but creates and sustains the interfaces immobilized at the

Figure 12.1 Scheme of an UF-RO membrane integrated desalination system [8].

j267

j 12 Integrated Membrane Processes

268

mouth of the pores; the separation process is based on the principles of phase equilibria. Membrane distillation (MD) is an example of membrane contactors applied to the concentration of aqueous solutions of nonvolatile solutes. In a MD process, a microporous hydrophobic membrane is in contact with an aqueous heated solution on the feed or retentate side. The hydrophobic nature of the membrane prevents the mass transfer in liquid phase and creates a vapor/liquid interface at the entrance of each pore. Here, volatile compounds (typically water) evaporate, diffuse and/or convect across the membrane, and are condensed and/or removed on the permeate or distillate side. With respect to RO process, MD does not suffer osmotic-pressure limitation and can be therefore employed when high permeate recovery factors or retentate concentrations are required. Membrane crystallization (MCr) has been recently proposed as one of the most interesting and promising extensions of the MD concept [13]. Evaporative mass transfer of volatile solvents through microporous hydrophobic membranes is employed in order to concentrate feed solutions above their saturation limit, thus obtaining a supersaturated environment where crystals may nucleate and grow. In addition, the presence of a polymeric membrane increases the probability of nucleation with respect to other locations in the system (heterogeneous nucleation) [14]. An integrated membrane desalination system based on NF, RO, and MD has been proposed [11]. In this system a NF unit has been used as pretreatment, while the MD contributed to concentrate the two brine streams from both NF and RO (Figure 12.2). The water production cost estimated for the integrated system was 0.92 $/m3 with a recovery factor of 76.2% [11]. Another integrated MF–NF–RO system having MD/MCr units operating on the NF and/or RO retentate, has also been proposed [10]. Five different integrated membrane system (IS) conﬁgurations have been considered: in the IS1 MF and NF are used as pretreatment to RO; in the IS2 a MCr unit operates on the NF retentate; in the IS3 a MCr works on the RO retentate; in the IS4 a MCr and a MD operates, respectively, on the NF retentate and on the RO retentate; ﬁnally in the IS5 two MCr operates both on the NF and RO retentate (Figure 12.3).

Figure 12.2 Scheme of a NF-RO-MD membrane integrated desalination system [11].

12.2 Integrated Membrane Processes for Water Desalination

Figure 12.3 Flow sheet of an integrated desalination system utilizing membrane crystallization units (IS5, Ref. 10).

In addition to the MF pretreatment, used to remove particulates and microorganisms from raw water, reducing membrane fouling in the successive steps, the NF step is able to reduce water hardness decreasing osmotic pressure of the RO feed, allowing operation at higher water recovery level. Moreover, when a MCr/MD is fed with the NF and/or RO retenates, salts crystals can grow in the high concentrated brines, or at least, a more concentrate solution is obtained, increasing the water recovery. Although the MCr/MD units introduce a thermal energy requirement, the waterrecovery factor is increased up to 92.8% for the IS5 without a relevant increase of the cost if waste thermal energy is already available for the process; at the same time brine ﬂow rate is signiﬁcantly reduced while the fresh water ﬂow rate is increased (Table 12.2) [10]. Moreover, the sale of salt crystals grown in highly concentrated brines (in particular MgSO47 H2O), might potentially reduce the overall desalination cost, thus conﬁrming the potential interest for the proposed approach. In fact, production of solid materials of high quality and controlled properties (speciﬁc polymorphs) with important added value, could transform the traditional brine-disposal cost in a potentially new proﬁtable market; reducing, moreover, the environmental problems of the brine disposal [10].

Table 12.2 Comparison of five integrated desalination systems (Data from Ref. [10].)

IS1 Fresh-water recovery (%) Water unit cost ($/m3) Brine ﬂow rate (m3 h1) Brine concentration (g L1) Fresh-water ﬂow rate (m3 h1) Fresh-water concentration (g L1) CaCO3 produced (kg m3) NaCl produced (kg m3) MgSO47 H2O produced (kg m3) a

49.2 0.46 531.9 68.02 517.6 0.270 — — —

IS2

IS3

IS4

IS5

71.6 0.68–0.55a 296.6 95.94 753.0 0.186 1.19 9.79 1.25

70.4 0.59–0.47a 309.9 76.53 739.6 0.189 0.12 16.86 0.00

88.6 0.74–0.55a 118.5 240.0 931.5 0.150 0.96 7.91 1.01

92.8 0.73–0.54a 74.6 214.4 974.9 0.143 1.00 20.35 0.96

If thermal energy is already available in the plant.

j269

j 12 Integrated Membrane Processes

270

An integrated water-treatment system designed to use different water sources and different treatment processes, including membrane processes, has been realized in Terneuzen (The Netherlands) [15]. Raw-water sources and treatments include: seawater and integrated membrane system to produce demineralized water; fresh water, and ion exchange to produce demineralized water; efﬂuent industrial wastewatertreatment plant (WWTP) and media ﬁltration to produce cooling tower supply water. The plant, operating from 2000, produces an 750 m3/h demineralized water, 650 m3/h cooling tower supply water and 1.050 m3/h ultrapure water [15]. In the integrated membrane system fed with seawater (Figure 12.4A) the pretreatment consists of two rotating microscreens of 150 mm used to remove the larger suspended particles from the water. Successively 8 MF units (polypropylene (PP) hollow membranes placed in a vertical position and operated according to the deadend principle) are used to remove the suspended solids completely, algae and to disinfect the water. The MF section is designed to ﬁlter 700–750 m3/h [15]. The permeate of the MF units is fed to two SWRO units equipped with highpressure pumps with an energy-recovery system (Pelton wheel). The SWRO units comprise 44 pressure vessels loaded with 6 Dow/FilmTec SW30 membranes each. The designed permeate capacity of each unit is 210 m3/h per unit with a recovery of 50–55%. Antiscalant is dosed to the feed stream of the SWRO with a concentration of 3 to 4 ppm [15]. While the SWRO permeate ﬂows into a reservoir and is successively fed to two BWRO units, the concentrate is directly discharged into the Westerschelde. The ﬁrst BWRO unit consists of 16 pressure vessels; the second BWRO unit consists of 6 pressure vessels. Each vessel contains 6 Dow/FilmTec BW30 membranes. The designed permeate capacity of each BWRO unit is 175 m3/h with a recovery of 85% [15]. The permeate of the BWRO (conductivity of 10–15 mS/cm) is mixed with the demineralized water produced by the ion-exchange process and consecutively

Figure 12.4 Process flow diagram of the water-treatment plant fed with seawater (A, period 2000–2006) and now operating on treated secondary communal wastewater (B, from 2007) [15].

12.3 Integrated Membrane Process for Wastewater Treatment

supplied at the Dow plant. The concentrate of the BWRO is mixed with the pretreated seawater and fed to the SWRO. In consideration of some operational difﬁculties, like biofouling and corrosion, directly related to the feed-water source, more reliable water sources have been evaluated by the Evides Industriewater that owns and operates this plant, like anaerobic groundwater and sweet tertiary wastewater. From the 2007 treated secondary communal wastewater (efﬂuent from the communal wastewater treatment plant in the city of Terneuzen) is used as the new feed-water source [16]. The integrated membrane system was reengineered (Figure 12.4B). The MF pretreatment section was not subject to hardware changes, but operated in a different regime [16]. The two SWRO units have been reengineered to low-pressure RO systems (EWRO array: 28–16 with membranes Dow Filmtec BW30-FR). The two BWRO units remained unchanged. The designed permeate capacity of the BWRO is 150–175 m3/h per unit operating at a recovery of 85%. The permeate of the BWRO has a conductivity of 10 mS/cm and is mixed with demineralized water originating from the ion-exchange process and consecutively supplied to Dow [15]. The concentrate of the integrated membrane system is mixed and discharged into the Westerschelde.

12.3 Integrated Membrane Process for Wastewater Treatment

As reported in the previous example, integrated membrane processes represent a viable solution also for wastewater treatment. Comparing three different treatments system (Figure 12.5): traditional treatment using coagulation/ﬂocculation, sand ﬁltration, physicochemical softening, activated carbon adsorption, and disinfection (A); spiral-wound nanoﬁltration with ultraﬁltration pretreatment followed by marble ﬁltration and disinfection (B); and direct capillary nanoﬁltration with only a limited pretreatment and post-treatment by marble ﬁltration and disinfection (C), the solution B and C showed better performance for quality and public health and operational aspects than the solution A [17]. Taking into account economical aspects and environment impact criteria the process C was more advantageous than B; however; the traditional treatment A resulted to be more advantageous for economical aspects [17]. More progresses can be anticipated in the near future by promoting the integration of different membrane operations, including MCs and membrane bioreactor (MBRs), also for wastewater treatment. MBR are already considered by the European Union as one of the best available technologies for municipal and industrial wastewater treatment. In a MBR, biological treatment is integrated with membrane ﬁltration, providing an effective alternative to conventional processes such as activated sludge for municipal and industrial efﬂuents [18].

j271

j 12 Integrated Membrane Processes

272

Figure 12.5 Scheme of three different drinking-water production plants: traditional (A); using ultrafiltration pretreatment and spiral-wound nanofiltration (B) and using capillary nanofiltration (C) [17].

Membrane bioreactors are composed of two fundamentals parts, the biological reactor in which the reaction occurs (active sludge containing purifying bacteria) and the membrane module for the separation of the different compounds. Membrane bioreactors can be classiﬁed into two main groups according to their conﬁguration. The ﬁrst group is commonly known as recirculated or external MBR and involves the recirculation of the solution through a membrane module that is outside the bioreactor. Both inner-skin and outer-skin membranes can be used in this

12.3 Integrated Membrane Process for Wastewater Treatment

application. The second conﬁguration is the integrated or submerged MBR that involves outer skin membranes internal to the bioreactor. In the external MBR the driving force is due to a difference in transmembrane pressure obtained by high-pressure crossﬂow along the membrane surface. In the submerged MBR the driving force is achieved by pressurizing the bioreactor or creating negative pressure on the permeate side. A diffuser is usually placed directly beneath the membrane module to facilitate scouring on the ﬁltration surface. Aeration and mixing are also achieved by the same unit. The ﬁrst example of an MBR operated with tubular membranes placed in external recirculation loops. However, the use of recirculation loops leads to relative high energy costs. In addition, the high shear stresses in the tubes and recirculation pumps can contribute to the loss of biological activity of the system [19]. Submerged MBRs are alternative systems to overcome these limits. This operating mode limits the energy consumption associated with the recirculation cost [20]. Moreover, the use of submerged membranes allow operation at low transmembrane pressures. This makes MBRs well suited to relatively large-scale applications. Submerged MBRs were ﬁrst introduced for decentralized sanitation in the US and in building-water reuse in Japan, and are now widely applied in different sectors. One of the largest membrane bioreator unit in the world was recently built in Porto Marghera (Venice, Italy) in order to extract remaining pollutants in tertiary water prior to disposal into the Venetian Lagoon [21]. The ultraﬁltration unit, containing submerged polyvinylideneﬂuoride (PVDF) hollow-ﬁber membranes (ZeeWeed by Zenon), is designed to treat 1600 m3/h of wastewater with a COD/h of 445 kg and the suspended matter of the treated water is <1 mg/L [21]. There are two interconnected UF lines, each line contains 4 unit composed by 9 ZeeWeed modules and the total membrane area is 100 000 m2 [21]. The application of membrane processes in the treatment of aqueous efﬂuents offers very interesting potentials also for the leather industry [22], traditionally considered one of the most polluting industries, being generally characterized by a low technological level of its operations. Wastes coming from the processing cycles of the leather are characterized by a high level of organic and inorganic pollutants, originating from natural skins or introduced during the treatment operations. The application of membrane-separation processes in the treatment of wastewater of the leather industry can give a reduction of the environmental impact, a simpliﬁcation of cleaning-up procedures of aqueous efﬂuents, an easy re-use of sludge, a decrease of disposal costs, and a saving of chemicals, water, and energy [22]. The separation operations of the leather cycle can be combined with or modiﬁed by membrane processes such as MF, UF, NF, and RO, showing in many cases a signiﬁcant improvement from an energetic point of view [22]. Experiments carried out in pilot-scale plants on liming, degreasing, and chromium tannage show the possibility to realize more efﬁcient operations for recovering by-products (e.g., proteins and fats) and chemicals (e.g., chromium) [22]. Recovering chromium salts from spent tanning liquors has been carried out with an integrated process based on preliminary UF followed by NF [23].

j273

j 12 Integrated Membrane Processes

274

The UF pretreatment (spiral-wound membrane module with a molecular weight cutoff of 50–100 kDa) allows a reduction of suspended solids (84%) and fat substances (71%) [23]. The NF step (spiral-wound membrane module with a molecular weight cutoff of 150–300 Da) gives a rejection 97% of the chromium and 98% of the sulfates in the feed stream, whereas the organic matter retention was 51% [23]. The concentrated chromium solutions obtained (ﬁnal value of about 10 g/L) was used in chrome tannage experiments performed on pickled sheepskins. The physical properties of these samples were compared with those of control skins treated with the conventional method using basic chromium sulfate containing 26% Cr2O3, observing improved characteristics of the ﬁrst ones (same hydrothermal stability grain crack and bursting strength values higher) [23]. A higher percentage of chromium recovery was obtained in the tanning operation performed with the NF retentate. In this process, the reuse of the NF permeate in the pickling step is also possible, adjusting the chloride concentration through the addition of NaCl, because the 58% of the chlorides is recovered in the permeate [23]. Alternatively, this solution can be used directly in the chrome tannage step according to a one-bath process in which skins are pickled at a pH of 3 or lower and then chrome tanning materials are introduced and the pH is raised (Figure 12.6). The main advantages of this integrated processes are: reduction of the environmental impacts, improving the leather quality, reducing the wastewater discharge and pollution load, saving of chemicals and water, allowing the reuse of sludges and decreasing disposal costs.

12.4 Integrated Membrane System for Fruit-Juices Industry

Integrated membrane processes are today proposed also in the dairy, food, and fruitjuices industries. The RO potentialities as a concentration technique to remove water from fruit juices for the production of high-quality fruit-juice concentrate are well known [24]. The most relevant advantages of the RO process over traditional evaporation are in the reduced thermal damage of the product, increase of aroma retention, and lower energy consumption, since the process is carried out at low temperature. Disadvantages of RO come from its inability to reach the high concentration of juices typically obtained by evaporation, because of osmotic-pressure limitation. As a consequence concentrations larger than 25–30 Brix total soluble solid content (TSS) cannot be reached with a single-stage RO [25]. However, technological advances related to the development of new membranes operations and innovative strategies of process design, have partially overcome this limitation. Membrane distillation, for example, is not subject to osmotic-pressure limitation and can be therefore employed in integrated systems when high permeate recovery factors or retentate concentrations are requested.

12.4 Integrated Membrane System for Fruit-Juices Industry

Figure 12.6 Scheme of an integrated UF/NF system for the recovery of chromium from spent tanning effluents [23].

Moreover, MD typically works at temperatures underneath the boiling point of the solution, for this reason it is appropriate for concentrating thermolabile solutes, as in the case of fruit juices. The driving force, that is, a partial pressure gradient, is obtained by a temperature gradient between the feed and the distillate side. In the osmotic distillation (OD), a partial pressure gradient is activated by a difference in concentration. As a consequence, OD can proceed at ambient temperature and ﬂavor and fragrance compounds can be more conveniently preserved, than in thermally activated concentration processes. Because of the high content of suspended solids and pectins in fruit juices, the use of a MF or UF pretreatment before the RO unit is also able to reduce the viscosity of the feed stream, increasing the transmembrane ﬂux. In this logic, a new integrated membrane process for the production of concentrated blood orange juice has been proposed as an alternative to thermal evaporation (Figure 12.7) [26]. Freshly squeezed juice is initially clariﬁed by UF, then it is concentrated by RO up to about 25 Brix and ﬁnally by an OD step to a ﬁnal concentration of about 60 Brix [26]. The comparison in terms of preservation of the total antioxidant activity and of the bioactive antioxidant components of the juice (ascorbic acid, anthocyanins, hydroxycinnamic acids, ﬂavanones) demonstrates that this integrated membrane

j275

j 12 Integrated Membrane Processes

276

Figure 12.7 Scheme of an integrated membrane process for fruit-juice concentration [26].

process is a valid alternative to the conventional thermal process to obtain highquality concentrated juice with a reduction in thermal damage and energy consumption. In the thermally concentrated juice the total antioxidant activity (TAA) is reduced by about 26%; ascorbic acid 30%, anthocyanins 36% with respect to the fresh squeezed juice. In the membrane processes the TAA reduction was about 15, ascorbic acid 15% and anthocyanins 20% [26]. A similar integrated system has been employed to produce highly concentrated black currant juice. The main difference is the use of MF as pretreatment for the disinfection of the raw juice. The concentrated juice (63–72 Brix) has more than three times higher anthocyanin content, than the raw juice (15–18 Brix) and a good taste [27]. An UF/OD integrated system has been used for the concentration of the cactus pear juice [28]. This juice is characterized by a high micro-organism load and it is typically clariﬁed by thermal treatment (>110 C) that signiﬁcantly changes the original color, ﬂavor, and taste. UF operations are able to produce a juice with a good microbiological stability, preserving organoleptic properties. The clariﬁed juice is then concentrated by OD up to about 60 Brix, maintaining its nutraceutical characteristics [28].

12.5 Integrated Membrane Processes in Chemical Production

Besides previously described examples of integrated membrane systems and much more reported in the literature, including applications in gas separation and the petrochemical industry [29], a special case of integrated or hybrid membrane systems, with a lot of interest in the logic of the sustainable growth, is represented by the catalytic membranes reactors (CMRs).

12.5 Integrated Membrane Processes in Chemical Production

Catalysis is today one of chemistrys most important and powerful technologies. Currently, about 90% of chemical manufacturing processes and more than 20% of all industrial products involve catalytic steps [30]. Catalytic reactions are intensively used in chemical industry, energy conversion, wastewater treatment and in many other processes. In the perspective to realize a sustainable growth, the application of membrane technology in integrated catalytic processes is very promising [31, 32]. Among the different heterogenization strategies, the entrapping of catalysts in membranes or, in general, the use of a catalyst conﬁned by a membrane in the reactor, offers new possibility for the design of new catalytic processes. Because there are many different ways to combine a catalyst with a membrane, there are numerous possible classiﬁcations of the CMRs. However, one of the most useful classiﬁcations is based on the role of the membrane in the catalytic process: we have a catalytically active membrane if the membrane has itself catalytic properties (the membrane is functionalized with a catalyst inside or on the surface, or the material used to prepare the membrane is intrinsically catalytic); otherwise if the only function of the membrane is a separation process (retention of the catalyst in reactor and/or removal of products and/or dosing of reagents) we have a catalytically passive membrane. The process carried out with the second type of membrane is also known as membrane-assisted catalysis (a complete description of the different CMRs conﬁgurations will be presented in a speciﬁc chapter). CMRs can offer viable solutions to the main drawback of homogeneous catalysis: catalyst recycling. In addition, the membrane can actively take part in the reactive processes by controlling the concentration proﬁles thanks to the possibility to have membranes with well-deﬁned properties by the modulation of the membrane material and structure. Moreover, the selective transport properties of the membranes can be used to shift the equilibrium conversion by the removal of one product from the reaction mixture (e.g., hydrogen in dehydrogenation reactions), or to increase the reaction selectivity by the controlled supply of the reagents (e.g., oxygen for partial oxidation reactions). The membrane can also deﬁne the reaction volume, for example, by providing a contacting zone for two immiscible phases, as in phase-transfer catalysis, excluding polluting solvents and reducing the environmental impact of the process. In numerous cases, membrane-separation processes operate at much lower temperature, especially when compared with thermal processes such as reactive distillation. As a consequence they might provide a solution for the limited thermal stability of either catalyst or products. Furthermore, by membrane-separation processes is possible also to separate nonvolatile components. The downstream processing of the products can be substantially facilitated when they are removed from the reaction mixture by means of a membrane [31]. In some cases, the heat dissipated in an exothermic reaction can be used in an endothermic reaction taking place at the opposite side of the membrane. Typical examples are hydrogenation/dehydrogenation reactions carried out by palladium or Pd-alloy membranes characterized by a 100% theoretical selectivity towards the hydrogen.

j277

j 12 Integrated Membrane Processes

278

Figure 12.8 Scheme of the combination of a dehydrogenation reaction (endothermic) on one surface of a Pd-based catalytic membrane and a hydrogenation reaction (exothermic) by the diffused hydrogen on the other membrane surface.

Combination of the dehydrogenation on one surface of the Pd-based catalytic active membrane and hydrogenation by the diffused hydrogen on the other surface has been proposed [33]. Thanks to the excellent thermal conductivity of these membranes, the heat released by the hydrogenation can be utilized to drive the endothermic dehydrogenation (Figure 12.8). The use of a mixed oxygen ion–electronic conductor membrane for oxygen separation with direct reforming of methane, followed by the use of a mixed protonic–electronic membrane conductor for hydrogen extraction has also been proposed in the literature [34]. The products are thus pure hydrogen and synthesis gas with reduced hydrogen content, the latter suitable, for example, in the Fisher–Tropsch synthesis of methanol [34]. Improved selectivity in the liquid-phase oligomerization of i-butene by extraction of a primary product (i-octene C8) in a zeolite membrane reactor (acid resin catalyst bed located on the membrane tube side) with respect to a conventional ﬁxed-bed reactor has been reported [35]. The MFI (silicalite) membrane selectively removes the C8 product from the reaction environment, thus reducing the formation of other unwanted byproducts. Another interesting example is the isobutane (iC4) dehydrogenation carried out in an extractor-type zeolite CMR (including a Pt-based ﬁxed-bed catalyst) in which the removal of the hydrogen allows the equilibrium limitations to be overcome [36]. Catalytic reactions can be combined in membrane-assisted integrated catalytic processes with practically all the membrane unit operations available today. Many examples of integration of membrane contactors, pervaporation, gas separation, nanoﬁltration, microﬁltration, and ultraﬁltration operations together with catalytic reactions, have been proposed in the literature. One of the most investigated ﬁelds is the pervaporation-assisted catalysis applied to equilibrium-limited reactions.

12.5 Integrated Membrane Processes in Chemical Production

PV is today considered as an interesting alternative for the separation of liquid mixtures that are difﬁcult or not possible to separate by conventional distillation methods. PV-assisted catalysis in comparison with reactive distillation has many advantages: the separation efﬁciency is not limited by relative volatility as in distillation; in pervaporation only a fraction of the feed is forced to permeate through the membrane and undergoes the liquid- to vapor-phase change and, as a consequence, energy consumption is generally lower compared to distillation. In the PV-assisted catalysis, pervaporation is usually used to extract continuously one of the formed products in order to improve conversion of the reactants or to increase reaction selectivity. By far the most studied reactions combined with pervaporation is esteriﬁcation. It is a typical example of an equilibrium-limited reaction with industrial relevance. The esteriﬁcation of acetic acid with ethanol has been investigated using zeolite membranes grown hydrothermally on the surface of a porous cylindrical alumina support (the catalyst used was a cation exchange resin) [37]. The conversion exceeded the equilibrium limit, by the selective removal through the membrane of water and reached to almost 100% within 8 h [37]. Nanoﬁltration-coupled catalysis was also widely used for catalyst compartmentalization in CMRs. A NF step has been used for the arylation of oleﬁns using as catalyst Pd(OAc)2(PPh3)2 with P(o-tolyl)3 as stabilizing agent, obtaining catalyst recycling, preventing metal contamination of the products and increasing reactor productivity [38]. The coupling of photocatalysis and polymeric membranes has been carried out using TiO2 as photocatalyst compartmentalized in the reactor by a membrane [39]. Various types of commercial membranes (ranging from UF to NF) and reactor conﬁgurations have been investigated [39]. The conﬁgurations with irradiation of the recirculation tank and catalyst in suspension conﬁned by means of the membrane, has been reported as the more promising for the 4-nitrophenol mineralization [39] The membrane function in this case is the conﬁning of the photocatalyst and maintaining the pollutants in the reaction environment until their complete mineralization. Photocatalysts have been also successfully heterogenized in polymeric (catalytically active) membranes. Novel photocatalytic membranes have been prepared by the heterogenization in PVDF membrane of the decatungstate (W10O324), a polyanionic metal-oxide cluster used as photocatalysts for oxidation reactions. Decatungstate exhibits especially interesting properties for the photocatalytic detoxiﬁcation of wastewater since its absorption spectrum (lmax ¼ 324 nm) partially overlaps the UV solar emission spectrum opening the potential route for an environmentally benign solar-photoassisted application [40]. However, decatungstate has also some relevant limitations: it is characterized by low quantum yields, small surface area, poor selectivity and limited stability at pH higher than 2.5 [41].

j279

j 12 Integrated Membrane Processes

280

Membrane technology could offer interesting possibilities in order to overcome these limitations and to improve the advantages of catalysis mediated by the decatungstate by: the multiturnover recycling associated to heterogeneous supports, the selectivity tuning as a function of the substrate afﬁnity towards the membrane, the effect of the polymeric microenvironment on catalyst stability and activity. Decatungstate, in the form of a lipophilic tetrabutilamonium salt ((n-C4H9N)4 W10O32), has been homogeneously dispersed in porous membranes made of PVDF (PVDF-W10). Solid-state characterization techniques conﬁrmed that catalyst structure and spectroscopic properties of decatungstate have been preserved once immobilized within the membranes [42–44]. These catalytic membranes have been successfully used in a photocatalytic membrane reactor for the mineralization of the phenol, one of the main organic pollutants in wastewater, demonstrating that these catalytic membranes are stable and recyclable [42]. The rate of phenol degradation as a function of catalyst loading in PVDF membranes and the effect of transmembrane pressure that inﬂuences the contact time substrate/catalyst, have been investigated. The rate of phenol degradation catalyzed by decatungstate in homogeneous phase and in heterogeneous phase (PVDF-W10 membrane) was similar; however, when the catalyst is immobilized in the polymeric membranes a higher mineralization degree of the phenol was observed [42]. The high mineralization activity of the PVDF-W10 membrane in comparison to the homogeneous catalyst can be ascribed to the selective absorption of the organic substrate from water on the hydrophobic PVDF polymer membrane that increases the effective phenol concentration around the catalytic sites. Moreover, the polymeric hydrophobic environment protects the decatungstate from the conversion over longer time to a less-reactive isomer that has a maximum absorption at a wavelength of 280 nm. The possibility of linking a catalyst only on the external surface of PVDF membranes modiﬁed using plasma treatments has also been investigated [45, 46] Photocatalytic membranes were prepared from the self-assembly of phosphotungstic acid (H3PW12O40, a polyoxometallate able to promote photo-oxidation reactions or acid-catalyzed reactions) on the surface of PVDF membranes modiﬁed by an Ar/NH3 plasma discharge [45, 46]. This new method is very versatile, and can be easily extended to the heterogenization of other catalysts on membrane surfaces. Decatungstate has also been heterogenized by phase-inversion techniques in membranes made of Hyﬂon, an amorphous perﬂuoropolymer [47]. Because of the low afﬁnity of the tetrabutilamonium salt of decatungstate with this inert polymeric matrix, the formation of irregular catalyst aggregates were observed. However, it was possible to improve the polymer–catalyst afﬁnity by functionalizing the catalyst with a ﬂuorous-tagged decatungstate ([CF3(CF2)7(CH2)3]3CH3N)4W10O32 (indicated as RfN4W10) [47]. In this form, the catalyst was successfully and homogeneously dispersed as spherical clusters of uniform size in Hyﬂon membranes. The cationic amphiphile RfN þ groups induced the self-assembly of the surfactantencapsulated clusters (RfN þ groups capped on W10O324) that, during membrane

References

formation, caused the formation of supramolecular catalyst assemblies, which were then stabilized within the polymeric matrix. It was therefore possible to tune the decatungstate self-assembling process through the proper choice of the conditions used during membrane preparation. These Hyﬂon-based catalytic membranes have been used to catalyze the photooxidation of ethylbenzene (neat), showing superior catalytic performance with higher turnover number and better selectivity when compared to homogeneous catalysts [47]. When dispersed in the Hyﬂon matrix, the efﬁcacy of ﬂuoro-containing decatungstate depends on the speciﬁc electrochemical environment of the catalytic sites, the high solubility of O2 in the Hyﬂon matrix, and the selectivity of the perﬂuorinated polymeric material towards the ethylbenzene reagent and the products 2-phenylethanol and acetophenone.

12.6 Conclusions

Membrane unit operations are today largely used in many different applications for their higher efﬁciency in comparison with traditional separation systems. Moreover the integration of different membrane operations in the same unit, or in combination with conventional ones, offers important beneﬁts in terms of product quality, plant compactness, environmental impact, and energetic aspects. Microﬁltration, ultraﬁltration, and nanoﬁltration are becoming standard in feed pretreatment for water desalination, wastewater treatment and fruit-juice concentration. The availability of new membrane processes such as membrane contactors and catalytic membrane reactors, the progresses in membrane-fouling control and the development of new membranes with well-controlled structures and properties, are recognized as key factors for the design of alternative production systems. In any case, a more systematic analysis of all the possible advantages and limitations caused by the introduction of one or more membrane operations in a process, is necessary. This can be realized by considering speciﬁc indicators that allow the progress of industrial processes towards sustainability to be quantiﬁed and their impact on the environment, economy and society to be measured.

References 1 Koros, W.J. (2007) Journal of Membrane Science, 300, 1. 2 Charpentier, J.-C. (2007) Industrial & Engineering Chemistry Research, 46, 3465–3485.

3 Drioli, E. and Fontananova, E. (2004) Chemical Engineering Research & Design, 82 (A12), 1557–1562. 4 Raluy, G., Serra, L. and Uche, J. (2006) Energy, 31, 2025–2036.

j281

j 12 Integrated Membrane Processes

282

5 Frenkel, V. (2008) International Desalination and Water Reuse Quarterly, 17, 47–50. 6 Drioli, E., Curcio, E. and Fontananova, E. (2006) Mass Transfer Operation– Membrane Separations, in Chemical Engineering, in Encyclopedia of Life Support Systems (EOLSS), Developed under the Auspices of the UNESCO (eds J. Bridgwater, M. Molzahn and R. Pohorecki), Eolss Publishers, Oxford, UK, http://www. eolss.net. 7 Pearce, G.K. (2007) Desalination, 203, 286–295. 8 Xu, J., Ruan, G., Chu, X., Yao, Y., Su, B. and Gao, C. (2007) Desalination, 207, 216–226. 9 Drioli, E., Curcio, E., Criscuoli, A. and Di Proﬁo, G. (2004) Journal of Membrane Science, 239, 27–38. 10 Drioli, E., Curcio, E., Di Proﬁo, G., Macedonio, F. and Criscuoli, A. (2006) Chemical Engineering Research & Design, 84 (A3), 209–220. 11 El-Zanati, E. and El-Khatib, K.M. (2007) Desalination, 205, 15–25. 12 Drioli, E., Criscuoli, A. and Curcio, E. (2006) Membrane Contactors: Fundamentals, Applications and Potentialities, Membrane Science and Technology Series, vol. 11, Elsevier, Amsterdam, Boston. 13 Curcio, E., Di Proﬁo, G. and Drioli, E. (2002) Desalination, 145, 173–177. 14 Curcio, E., Fontananova, E., Di Proﬁo, G. and Drioli, E. (2006) The Journal of Physical Chemistry. B, 110 (25), 12438–12445. 15 van Agtmaal, J., Huiting, H., de Boks, P.A. and Paping, L.L.M.J. (2007) Desalination, 205, 26–37. 16 http://www.evides.nl/industriewater/ media/m260.pdf. 17 van der Bruggen, B., Verberk, J.Q.J.C. and Verhack, J. (2004) Water SA, 30, 413–419. 18 Drioli, E. and Giorno, L. (1999) Biocatalytic Membrane Reactors: Application in Biotechnology and the Pharmaceutical Industry, Taylor & Francis Publisher, Padstow, UK.

19 Brockmann, M. and Seyfried, C.F. (1997) Water Science and Technology, 35 (10), 173–181. 20 him, J.K., Yoo, I.-K. and Lee, Y.M. (2002) Process Biochemistry, 38, 279–285. 21 Vigiano, C. (2007) La Chimica e lIndustria, 5, 90–94. 22 Cassano, A., Molinari, R., Romano, M. and Drioli, E. (2001) Journal of Membrane Science, 181, 111–126. 23 Cassano, A., Della Pietra, L. and Drioli, E. (2007) Industrial & Engineering Chemistry Research, 46, 6825–6830. 24 Rao, M.A., Acree, T.E., Cooley, H.J. and Ennis, R.W. (1987) Journal of Food Science, 52, 375–378. 25 Jiao, B., Cassano, A. and Drioli, E. (2004) Journal of Food Engineering, 63, 303–324. 26 Galaverna, G., Di Silvestro, G., Cassano, A., Sforza, S., Dossena, A., Drioli, E. and Marchelli, R. (2008) Food Chemistry, 106, 1021–1030. 27 Kozak, A., Banv€olgyi, S., Vincze, I., Kiss, I., Bekassy-Molnar, E. and Vatai, G. (2008) Chemical Engineering and Processing, 47, 1171–1177. 28 Cassano, A., Conidi, C., Timpone, R., DAvella, M. and Drioli, E. (2007) Journal of Food Engineering, 80, 914–921. 29 Drioli, E. and Curcio, E. (2007) Journal of Chemical Technology and Biotechnology (Oxford, UK, 1986), 82, 223–227. 30 Zhao, X.S., Bao, X.Y., Guo, W. and Lee, F.Y. (2006) Materials Today, 9, 32–39. 31 Vankelecom, I.F.J. (2002) Chemical Reviews, 102, 3779–3810. 32 Miachon, S. and Dalmon, J.-A. (2004) Topics in Catalysis, 29, 59–65. 33 Gryaznov, V. (1999) Catalysis Today, 51, 391–395. 34 Norby, T. (1999) Solid State Ionics, 125, 1–11. 35 Piera, E., Tellez, C., Coronas, J., Menendez, M. and Santamarıa, J. (2001) Catalysis Today, 67, 127–138. 36 van Dyk, L., Miachon, S., Lorenzen, L., Torres, M., Fiaty, K. and Dalmon, J.-A. (2003) Catalysis Today, 82, 167–177.

References 37 Tanaka, K., Yoshikawa, R., Ying, C., Kita, H. and Okamoto, K. (2001) Catalysis Today, 67, 121–125. 38 Nair, D., Scarpello, J.T., White, L.S., Freitas dos Santos, L.M., Vankelecom, I.F.J. and Livingston, A.G. (2001) Tetrahedron Letters, 42, 8219–8222. 39 Molinari, R., Palmisano, L., Drioli, E. and Schiavello, M. (2002) Journal of Membrane Science, 206, 399–415. 40 Texier, J., Giannotti, C., Malato, S., Richter, C. and Delaire, J. (1999) Catalysis Today, 54, 297–307. 41 Mylonas, A. and Papaconstantinou, E. (1996) Polyhedron, 15, 3211–3217. 42 Fontananova, E., Drioli, E., Donato, L., Bonchio, M., Carraro, M. and Scorrano, G. (2006) Chinese Journal of Process Engineering, 6, 645–650.

43 Bonchio, M., Carraro, M., Scorrano, G., Fontananova, E. and Drioli, E. (2003) Advanced Synthesis & Catalysis, 345, 1119–1126. 44 Bonchio, M., Carraro, M., Gardan, M., Scorrano, G., Drioli, E. and Fontananova, E. (2006) Topics in Catalysis, 40, 133–140. 45 Fontananova, E., Donato, L., Drioli, E., Lopez, L., Favia, P. and dAgostino, R. (2006) Chemistry of Materials, 18, 1561–1568. 46 Lopez, L.C., Buonomenna, M.G., Fontananova, E., Iacoviello, G., Drioli, E., dAgostino, R. and Favia, P. (2006) Advanced Functional Materials, 16, 1417–1424. 47 Carraro, M., Gardan, M., Scorrano, G., Drioli, E., Fontananova, E. and Bonchio, M. (2006) Chemical Communications, 43, 4533–4535.

j283

Part Two Transformation This Part will be focused on the fundamentals and applications of membraneassisted transformation processes, i.e. membrane reactors. Two separate chapters are dedicated to the fundamentals of membrane reactors using traditional chemical catalysts at high temperature (>200 C) and catalysts of biological origin or biomimetic at low temperature (<100 C). Application of membrane reactors for degradation of organic compounds via photocatalysis, treatment of wastewater, production of bioactive high added value compounds, and biomedical treatments are discussed. The possibility of having membrane systems also as tools for a better design of chemical transformation is today becoming attractive and realistic. Catalytic membranes and membrane reactors are the subject of signiﬁcant research efforts at both academic and industrial levels. For biological applications, synthetic membranes provide an ideal support to catalyst immobilization due to their biomimic capacity; enzymes are retained in the reaction side, do not pollute the products and can be continuously reused. The catalytic action of enzymes is extremely efﬁcient, selective and highly stereospeciﬁc if compared with chemical catalysts; moreover, immobilization procedures have been proven to enhance the enzyme stability. In addition, membrane bioreactors are particularly attractive in terms of eco-compatibility, because they do not require additives, are able to operate at moderate temperature and pressure, and reduce the formation of by-products. Potential applications have been at the origin of important developments in various technology sectors, including genetic engineering of micro-organisms to produce speciﬁc enzymes, techniques for enzyme puriﬁcation, bioengineering techniques for enzyme immobilization, design of efﬁcient productive cycles. The development of catalytic membrane reactors for high-temperature applications became realistic only in the last few years with the development of high-temperatureresistant membranes. Due to the generally severe conditions of heterogeneous catalysis, most catalytic membrane reactors applications use inorganic membranes that can be dense or porous, inert or catalytically active. The scientiﬁc literature on catalytic membrane reactors is signiﬁcant today; however, practically no large-scale industrial applications have been reported so far because of the relatively high price of membrane units. However, current and future advancements in material engineering might signiﬁcantly reverse this trend. Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

The good H2 selectivity and permeability of the last-generation dense (Pd-based) and almost dense SiO2 membranes have been successfully exploited for a number of H2-consuming or -generating reactions; for some applications, the thermochemical instability of Pd membranes and the hydrothermal instability of silica remain the main problems to solve. Concerning O2-generating or -consuming reactions, the development of O2 permselective membranes with good ﬂuxes in the range of 400–700 C is still a challenge. Direct catalytic conversion of natural gas and light alkanes into oxygenates, fuels and higher hydrocarbons is currently one of the most strategic research topics for fundamental and industrial catalysis. The challenge to develop viable processes for the valorization of large reserves of natural gas and light alkanes along with the stringent need to innovate the conventional processes for the catalytic oxidation of hydrocarbons have resulted in a great research effort in this area since the 1970s. In this context, syngas production is a key step for the production of chemicals and fuels from natural gas. The current technology is based on steam or autothermal reforming, the partial oxidation with O2 being costly owing to the associated airseparation plant; for this application, oxygen-selective membranes integrated with the reactor represent an interesting alternative. Other innovative catalytic membrane reactors have been proposed for the direct conversion of natural gas into higher added value products, such as that based on the adoption of a proton conductor ceramic membrane that ensures H2 removal in the aromatization of CH4.

j287

13 Fundamental of Chemical Membrane Reactors Giuseppe Barbieri and Francesco Scura

13.1 Introduction

Membrane reactors (MRs) are an interesting alternative to traditional reactors (TRs) owing to their characteristic of product separation during the reaction progress. The simultaneous separation shows some advantages related to the process of both permeate and retentate downstreams and on the reaction (rate) itself. In fact, the load of the downstream separation is signiﬁcantly lower because both (permeate and retentate) streams leaving the MR are concentrated in more and fewer permeable species, respectively. In addition, separation/puriﬁcation is not required in the special case of pure permeate. From the reaction point of view, the product removal (1) reduces the ﬂow rate of the reactant stream, in the meantime increasing the residence time; (2) increases the reactant concentration and hence the forward reaction rate; (3) reduces product concentration, reducing the reverse reaction rate. The rate-determining steps of the reaction could change because, even though the species present are the same, they have a very different concentration with respect to a TR. The overall effect is a higher net reaction rate and residence time, both improve the performance (conversion, selectivity, yield, etc.) of the MRs itself. This implies a large reduction of the reaction volume [1, 2] and catalyst amount in the case of catalytic reaction. The improved performance indicates also the possibility of operating in conditions (e.g., of temperature and pressure) better than those used in the present industrial processes. The advantages of MR (volume reduction, improved conversion, two more valuable streams instead of one, etc.) pursue the logic of process intensiﬁcation [3, 4] (Figure 13.1) today the better strategy for sustainable growth compatible with a high-quality lifestyle, mitigating greenhouse-gas emissions, reducing investment and operating costs of chemical plants. It is a new design philosophy aimed at signiﬁcant reduction (by a factor of 10, 100 or more) in plant volume also because process equipment, piping, and so on are 20% of the plant costs.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 13 Fundamental of Chemical Membrane Reactors

288

Figure 13.1 Process intensification strategy: the trend required to variables.

Gas-phase reactions and in particular hydrogenation/dehydrogenations (e.g., hydrogen production, upgrade and clean-up) will be considered as an example in this chapter since they have been widely studied [5–34] and some investigations are also in progress both experimentally and by modeling [35] as well as the deﬁnition of the upper limit of conversion from a thermodynamic point of view is of speciﬁc interest [36, 37]. Porous and dense, ceramic or metallic membranes have only a separation function; they do not have catalytic properties. The scheme of a catalytic MR as shown in Figure 13.2 presents a tube-in-tube conﬁguration, where the inner tube is the selective membrane. The permeating species are collected inside the core of the inner tube. The permeation can be driven by a sweep gas; the use of feed pressure for permeation promoting is better because the permeate stream is concentrated in the permeating species and an extra separation of the sweep component is not required. Both (annular space or tube core) volumes can be used for reaction but the choice depends also on the energy transport [38]. If the reaction is energy intensive for example, methane steam reforming, the annular space gives an overall heat-exchange coefﬁcient higher than that shown by the same MR geometry with the catalyst packed inside the membrane.

Figure 13.2 Scheme of a tube-in-tube MR.

13.2 Membranes

Figure 13.3 Scheme of a catalytic membrane with a cylindrical geometry (tubular or hollow fiber).

Catalytic membranes are also very interesting. The reactants ﬂowing through the membrane pores pass at a short distance from the catalytic sites and the probability of an interaction is much higher (see Figure 13.3). A catalytic membrane leaves very little room to by-pass: the reactant stream has to ﬂow inside the membrane. In addition, very high conversion might also be achieved in the puriﬁcation process [39] operated by reaction when the component to be removed has a lower concentration with respect to the other desired species. Table 13.1 reports, as an example, a list of some hydrogenation and dehydrogenation reactions investigated in MRs. It also reports reactions of hydrogen production and its upgrade (MSR, WGS, etc.), because they are widely studied owing to their deep interest in the energy ﬁeld, hydrogen being one of the most used energy carriers. All these reactions are typically operated at a temperature higher than 200 C. In addition, reactions carried out using polymeric membranes (at a temperature lower than 100 C) are cited just as an example of the use of polymeric membranes in ﬁne-chemicals production.

13.2 Membranes

A fundamental innovative aspect of chemical MRs is the separation operated by means of the membrane. In fact, the mass-balance equations have to take into account the mass transfer due to the permeation through the membrane; the other terms being the same present in mass-balance equations of TRs. Figure 13.4 introducing reaction paths for MSR at different values of hydrogen permeance shows a very large difference with the permeance and also with the path of a TR.

j289

j 13 Fundamental of Chemical Membrane Reactors

290

Table 13.1 Examples of reactions investigated in catalytic MRs and catalytic membranes.

Reaction

Membrane

Hydrogenation of . . . butanes butadiene ethylene to ethane

dense dense dense; porous

Pd-Sb Pd Pd; Al2O3

Dehydrogenation of . . . ethane to ethylene ethylbenzene to styrene 1-butene to butadiene iso-butene n-butane methanol

dense; porous dense dense dense porous porous

Pd-Ag; Al2O3 Pd-Ag Pd Silica g-Al2O3 g-Al2O3

Other Methane steam reforming, MSR Methane dry reforming Partial oxidation of methane, POM

dense dense dense

Water gas shift, WGS CO clean-up Decomposition of H2S Oxidation of secondary ammine (liquid phase) Photo-oxidation of n-pentanol (liquid phase)

dense porous porous porous, dense porous, dense

Pd-alloy Pd- alloy Pd-based, perovskite Pd- alloy zeolite g-Al2O3 polymeric Polymeric

Therefore, the peculiarity of the permeation mechanisms will be discussed because it is introduced later in the balance equations. Dense and porous, metallic, polymeric, zeolitic, of zirconia, alumina, and so on membranes were used and are in use for MR investigations (Table 13.1). The choice depends on the reaction and operating conditions as well as on the separation desired. The peculiarities of some membranes that are important for the subsequent discussion are recalled here. Pd-based membranes are the most used dense membranes because they have inﬁnite hydrogen selectivity. The permeation follows Sieverts law (13.1) that identiﬁes the hydrogen diffusion in the metal bulk as the ratedetermining step and the difference of the partial pressure of the hydrogen on both membrane sides as the driving force. The permeation is due to several elementary steps [40, 41]. Hydrogen permeance (Figure 13.5) can be reduced by interaction between the membrane surface and some chemical species such as CO [42]. Perovskite membranes transport (13.2) the oxygen [43] through a dense layer of La, Fe, Co, Sr, and so on oxides; the operating temperature is very high (800–900 C). These membranes were used in partial oxidation reactions [44] where diffused oxygen feed is a signiﬁcant improvement for driving the reaction selectively towards the desired products. Polymeric dense membranes are less selective than Pd-alloy or perovskitic membranes, the permeation is due to the diffusion (13.3) of the solubilized species in

13.2 Membranes

Figure 13.4 Effect of the permeance (ranging from 0.01 to 10 times p, used as a reference value of the permeance) on the MR performance. Two configurations (reaction in the annular space or in the core of the tube) are considered. The curves indicated with p and 10 p (tube) and 0.1 p (tube) are the reaction paths when the catalyst is packed inside the tube and thus

the reaction takes place inside the membrane. TR is the TR path. The dashed curve represents the conditions of 60% of the MREC; MREC being the curve B-C10TFeed ¼ TSweep ¼ 500 C, PReaction ¼ PPermeation ¼ 100 kPa, H2O/CH4 feed Feed molar ratio ¼ 3, QCH4 ¼ 400 cm3 ðSTPÞ min1 ; Feed Q Sweep ¼ 10 QCH4 , UShell ¼ 227 W m2 K1, UMembrane ¼ 2.4 W m2 K1.

Figure 13.5 H2 flux as function of Sieverts driving force. Scheme summarizing dilution and inhibition effects owing to inert and inhibitor species present in mixture with hydrogen.

j291

j 13 Fundamental of Chemical Membrane Reactors

292

the polymeric bulk. These membranes were also used for catalyst heterogenization for, for example, photo-oxidation and secondary ammine oxidations [45, 46]. Knudsen is a transport mechanism (13.4) of species in porous structures. Zeolitic membranes were used in hydrogen puriﬁcation of stream containing CO [39]. Sieverts law Permeating

JH2

¼

QH2 eðEp=RT Þ Thickness

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Reaction side Permeation side PH PH 2 2

with CO inhibition effect kCO PCO QH2 eð-Ep=RT Þ Permeating ¼ 1aðTÞ JH2 1 þ kCO PCO Thickness

ð13:2Þ

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Reaction side Permeation side PH PH 2 2

Perovskite membranes Permeating JO2

ð13:1Þ

kr ¼ kp

1

2 Thickness 1 þ þ Permeation side Reaction side D kexternal kexternal V qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ side side POReaction POPermeation 2 2

!1

Solution-diffusion Permeating

Ji

¼

Solubilityi Diffusivityi Reaction side Permeation side Pi Pi Thickness

Knudsen flux Permeating

Ji

¼ dpore

e t

rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Reaction side Permeation side Pi Pi 1 Thickness 3 RTMi

ð13:3Þ

ð13:4Þ

Some of the variables that are important for the subsequent discussion are recalled here. The membrane properties are related to the mass transport of the different chemical species through the membrane itself or its separating layer (for an asymmetric or multilayer membrane). Permeability and selectivity were deﬁned for the mass transport by permeation; both depend on the membrane nature and morphology that impose the speciﬁc transport mechanism driving the permeation of which it is characteristic. Table 13.2 reports the permeability coefﬁcient, selectivity and permeating driving force of some permeation mechanisms.

(2)

(3)

(4)

Ion transport

Solution-diffusion

Knudsen

dpore

e t

rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 ðmol m1 s1 Pa1 Þ 3RTMi

Solubilityi Diffusivityi (mol m1 s1 Pa1)

ðmol m1 s1 Pa0:5 Þ

Thickness

1 2 Thickness 1 þ þ Permeation side side DV kReaction kexternal external

Q0 eðEp=RT Þ ðmol m1 s1 Pa0:5 Þ

(1)

Sieverts kr kp

Permeability

Transport mechanism !1

Table 13.2 Permeability, selectivity and permeating driving force of some transport mechanisms.

ðPiReaction PiPermeate Þ; Pa

ðPiReaction PiPermeate Þ; Pa

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ side P Permeation side ; Pa0:5 POReaction O2 2

Permeation driving force qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Reaction P Permeate ; Pa0:5 PH H2 2

Solubilityi Diffusivityi Solubilityj Diffusivityj sﬃﬃﬃﬃﬃﬃ Mj Mi

Inﬁnite to oxygen

Inﬁnite to hydrogen

Selectivity aij, -

13.2 Membranes

j293

j 13 Fundamental of Chemical Membrane Reactors

294

The permeability value of the speciﬁc species (ith) can be evaluated as the ratio of the permeating ﬂux (experimentally measurable) and the gradient of the permeating driving force: Permeabilityi ¼

Permeating fluxi ; dPi =dðThicknessÞ

mol m1 s1 Pa1

ð13:5Þ

When membrane thickness is unknown, the permeance, which is the ratio between the permeability and membrane thickness can be used. Permeancei ¼

Permeabilityi ; Thickness

mol m2 s1 Pa1

ð13:6Þ

The performance of a speciﬁc already produced membrane characterized by a deﬁned (known or unknown) thickness depends on permeance, which can be evaluated by means of a permeation measurement. The permeability is a property of the membrane material of the separating layer, whereas the permeance is also a property of the product membrane in its entirety considering thickness, the eventual support, defects, and so on. For this reason, the permeance will be used instead of permeability in subsequent mass-balance equations. The ratio of the permeability (or permeance) of two species (ith and jth) deﬁnes the selectivity: Selectivityi;j ¼

Permeabilityi Permeancei ¼ Permeabilityj Permeancej

ð13:7Þ

The permeation (mass transport through the membrane) generates a concentration gradient in the orthogonal direction to the membrane surface. Thus, the concentration of the faster permeating species (generally, it is desired in the permeate) reduces, and in the meantime the concentration of the less-permeating compound increases. Thus, the permeation of the desired species is reduced too. This phenomenon, already known as concentration polarization, affects the permeation as well as the system performance. It can be taken into account in the permeance considering the concentration-polarization coefﬁcient [47]. An onerous alternative solution is 2D mathematical models including radial diffusion [35].

13.3 Membrane Reactors 13.3.1 Mass Balance

The mass balances for all the species involved, for both the tube and shell side of the system shown in Figure 13.2 (shell-side feed conﬁguration, system with cylindrical symmetry) can be written as:

13.3 Membrane Reactors

Shell side with reaction, species ith reaction qcireaction 1 q reaction qNi;z rNi;r þ þ qt qz r qr

! ¼

NReactions X

ð13:8Þ

ui;j rj

j¼1

Tube-side only permeation, species ith permeation

qci

qt

permeation

1 q permeation qNi;z þ þ rNi;r qz r qr

! ¼0

ð13:9Þ

In particular, for the species A in the case of Fick diffusion of binary mixtures [48] the ﬂux is given: A ¼ xA ctotalvctotal DAB rxA N

ð13:10Þ

xA ctotalv being the convective contribute and (ctotalDAB!xA) the diffusive contribi could be ute. In a more general case, such as a multicomponent system, the ﬂuxN expressed by means of Maxwell–Stefan equations. However, also momentum balance (Equation (13.11), [48]) has to be solved for both i , strictly the shell and tube sides in addition to the mass balances (the molar ﬂux,N depends on the velocity ﬁeld, v). qrv g ¼ ½r frvv þ p qt

ð13:11Þ

A set of boundary and initial conditions (BCs and ICs) is necessary to solve the system of Equations 13.8, 13.9 and 13.11. The speciﬁc contribution of the permeation is expressed by means of a BC related to the membrane surface for both reaction (13.8) and permeation side (13.9). It is equal to the permeating ﬂux. Reaction side reaction Ni;r

Membrane 2

r¼OD

¼ Ji ¼ Permeancei Driving forcei

ð13:12Þ

Permeation side reaction Ni;r

Membrane r¼ID 2

¼ Permeancei Driving forcei

ODMembrane IDMembrane

ð13:13Þ

A 2D model requires a signiﬁcant computational effort. If the radial diffusion (Ni,r ¼ 0) can be considered negligible the balance equations are much simpler: the set of equations is of PDEs, 1D of the second order. Thus, the computational requirement is also reduced.

j295

j 13 Fundamental of Chemical Membrane Reactors

296

Shell (reaction zone) qPiReaction q2 PiReaction qNiReaction 1 ¼ D i RT Reaction RT Reaction qt qz2 qz NReactions X AMembrane Permeating ui;j rj Reaction Ji þ V j¼1 1

I:C:

PiReaction ðzÞ

B:C: 1; 2

¼

PiReaction; Initial

PiReaction z¼0 ¼ PiFeed ;

ð13:14Þ

qPiReaction ¼0 qz z¼L

Tube (permeation zone) 1 RT Permeation

qPiPermeation q2 PiPermeation qNiPermeation 1 Di ¼ Permeation qt qz2 qz RT þ

AMembrane Permeating J V Permeation i

I:C: PiPermeation ðzÞ ¼ PiPermeation; Initial B:C:1; 2

Sweep PiPermeation z¼0 ¼ Pi ;

qPiPermeation ¼0 qz z¼L ð13:15Þ

A 1D model does not require any BC in the radial direction and the term related to the Permeating permeation ðJi Þ is now directly present in the mass-balance equations. Any 1D model for a tubular MR, not considering any gradient along the permeation direction, could lead to too strong an approximation. In this case, the concentration-polarization coefﬁcient can be introduced as a reducing factor of the permeance and hence the permeating ﬂux is consequently reduced. This solution allows the introduction of the concentration-polarization coefﬁcient for the actual permeance value and by integrating a 1D model good model results to be obtained. The determination/estimation of the concentration-polarization coefﬁcient remains an important task to be solved, in this case. The polarization in the gas-phase reaction is not so important; thus, it can be neglected in most cases. It has to be considered only if the permeation is faster than product formation and its incoming ﬂow [47]. 13.3.2 Energy Balance

Usually, heat is developed during chemical reactions, changing the temperature along the reactor. Any temperature variation means variations of kinetics (reactions rate) and, speciﬁcally for MRs, permeation. Therefore, the energy balance has to be considered as part of the equation set in addition to the mass balances. Below, the energy balance is written down for a 1D system: the same as used for writing the last mass-balance equations. The following equations contain the heat exchange between the two

13.3 Membrane Reactors

membrane sides and that transported by the permeated species, in addition to the typical terms of the TRs. The contribution owing to the permeated species is different from zero on the permeate side (it contributes to the increase in the temperature), but is null on the other side, because it leaves the reaction side at the same temperature. Figure 13.6 shows a very different behavior of MRs depending on heat-transfer coefﬁcients owing to different temperature proﬁles developed inside the MR. Nspecies X

Ci Cpi

i

Nspecies X qT Annulus qT Annulus q2 T Annulus Ni Cpi ¼ þ kz qt qz qz2 i¼1

Shell

U AShell Furnace U Membrane AMembrane Annulus ðT T Þ V Annulus V Annulus AMembrane ðT Annulus T Lumen Þ þ Y þ F Annulus V I:C: T Annulus t¼0 ¼ T Annulus; Initial B:C:1; 2

T Annulus z¼0 ¼ T Feed or T Sweep ;

Nspecies X

Ci Cpi

i

þ

qT Annulus qz

¼0

Nspecies X qT Lumen qT Lumen q2 T Lumen ¼ þ kz Ni Cpi qt qz qz2 i¼1

U Membrane AMembrane Annulus AMembrane ðT T Lumen Þ þ Y þ F Lumen Lumen V V

I:C: T Lumen jt¼0 ¼ T Lumen; Initial B:C:1; 2

ð13:16Þ

z¼L

T Lumen jz¼0 ¼ T Feed or T Sweep ;

ð13:17Þ

qT Lumen ¼0 qz z¼L

Temperature variation owing to enthalpy ﬂux associated with permeation ( 0 on reaction side F¼ Reaction side Permeation side Permeating Ji ðhiT hiT Þ on permeation side

ð13:18Þ

Heat produced by chemical reactions

Y¼

8 Reactions NX > > < rj ðDHj Þ on reaction side > > :

j¼1

ð13:19Þ

0 on permeation side

Analysis of the effect of permeation, temperature proﬁle and sweep gas will be proposed hereafter considering a steady-state MR modeled by a 1D, ﬁrst-order model. The model can be extracted from the mass and energy balance, Equations 13.14

j297

j 13 Fundamental of Chemical Membrane Reactors

298

Figure 13.6 Conversion versus reactor length for annular, luminal and adiabatic MRs. Inlet conditions: PReaction ¼ PPermeation ¼ 100 kPa, Sweep factor ¼ 10, H2O/CH4 Feed feed molar ratio ¼ 3 and QCH4 ¼ 200 cm3 ðSTPÞ min1 .

and 13.16, respectively, deleting the transient and second-order terms and the related ICs and BCs. Figure 13.6 shows the different (higher) conversion reached by an MR with respect to a TR. The MR length for a set conversion value depends strongly on the overall heat-exchange coefﬁcient (annular or luminal MR) for the MSR reaction (highly endothermic). When the catalyst is packed in the annular volume the energy is supplied faster for two reasons: (1) the energy transfer has to pass through the stainless steel shell and both gaseous ﬁlms adjacent to the shell itself; (2) the thermal resistance is not so high: it depends mainly on gaseous ﬁlms. On the contrary in the other case (catalyst packed in the core of the tube), the energy required by the reactions also has to cross the membrane. If the membrane is supported on porous alumina, as in this speciﬁc case, the overall heat-transfer coefﬁcient covering the supported membrane is very low. The reaction, requiring a lot of energy, cannot proceed as fast as the other (annular) case. In any case, both MR conﬁgurations achieved MREC (for a sufﬁciently long reactor length) and exceed the TR equilibrium conversion (TREC), the maximum values achievable, which is also reported. Only in the case of adiabatic condition does the MR show not a good performance because the temperature proﬁle goes down along the reactor length and the ﬁnal conversion is that of equilibrium at the corresponding temperature. The same behavior can be observed in another diagram type (Figure 13.4) in which the methane conversion is plotted against temperature. The annular MR shows, at any point, a temperature always higher than that showed by a tubular MR. This means a higher distance from the MREC (curve B-C10), lower reaction rate and permeance (owing to the temperature), and so on. The same ﬁgure also shows the effect owing to the permeation. A signiﬁcant reduction (100-fold) of permeance (for the annular MR) gives MR behavior close to that of a TR: low permeance leads the MR to approaches a TR. For a tubular MR this effect is not signiﬁcant because the

13.3 Membrane Reactors

Figure 13.7 Radial concentration profiles for an A ! B ! C reaction in the case of convection-diffusion-reaction in a catalytic hollow-fiber membrane.

rate-determining step is the energy supply. All the reaction paths reach the MREC value (point C10) at the furnace temperature. The permeance reduction could be due to a different chemical nature or thickness of the separating layer, or surface phenomena such as that discussed with regard to Figure 13.7. Another fundamental aspect of an MR is related to the permeation driving force. Any system with a permeance value different from zero gives permeation in the desired direction under a suitable driving force. It can be generated by means of an appropriate value of feed pressure or using a sweep gas. Figure 13.8 shows the MREC

Figure 13.8 Methane conversion as a function of the temperature at several sweep factors (solid lines). H2O/CH4 feed molar ratio ¼ 3, PReaction ¼ PPermeation ¼ 100 kPa. TR with inert: equilibrium of a TR when an inert stream with a flow rate equal to 10 times that of methane was

added to the feed. The points BI and CI are the final points of the TR and MRs obtained by simulation in adiabatic, isothermal, and nonisothermal reactors, respectively, with the following conditions: TReaction ¼ TPermeation ¼ 500 C.

j299

j 13 Fundamental of Chemical Membrane Reactors

300

Figure 13.9 MSR reaction. CH4 equilibrium conversion for both traditional and membrane reactors. I is the ratio of the sweep flow rate to the CH4 feed flow rate. H2O/CH4 feed molar ratio ¼ 3, permeate pressure ¼ 100 kPa.

as a function of the temperature at different values of the sweep factor. This new parameter is the ratio between the sweep-gas ﬂow rate and the reference reactant ﬂow rate; it is deﬁned in an analogous way to the feed molar ratio of a TR. The MREC trend is the same as that of a TR with the temperature: the temperature does not change any functionality of reaction. On increasing the sweep factor the permeate side has a higher removal capacity and hence the conversion is increased too. If the same sweep ﬂow is fed into a TR together with the reactants the conversion, represented by the dashed lines, even though higher than that of a TR, is signiﬁcantly lower than that shown by an MR. The effect of the feed pressure is shown in Figures 13.9 and 13.10 for equilibrium conversion. Any pressure increase produces an increase of the MREC. The nonequilibrium conversion goes in the same direction.

Figure 13.10 WGS reaction. CO equilibrium conversion for both traditional and membrane reactors. I is the ratio of the sweep flow rate to the CO feed flow rate. H2O/CO feed molar ratio ¼ 1, permeate pressure ¼ 100 kPa.

13.5 Thermodynamic Equilibrium in Pd-Alloy Membrane Reactor

13.4 Catalytic Membranes

Catalytic membranes are characterized by reaction and permeation at the same point inside the membrane itself. This very interesting case is depicted in Figure 13.3 for a cylindrical (hollow-ﬁber or tubular) membrane. Any cross-section for the mass transport has a cylindrical shape (circumference in Figure 13.3). Figure 13.3 shows the membrane section along the permeation and reaction direction, the ﬁgure also shows an indicative proﬁle of a reactant and product. The mass-transport equations for this system, focusing attention inside the membrane and neglecting the axial proﬁle along the ﬁber, are: NReactions X qci qci 1 qci qci þ vr ðrÞ ¼ Deffective þ 2 e ui;j rj qt qr qr r qr j¼1 I:C: t ¼ 0 ci ¼ 0;

e

B:C:1

r ¼ ODMembrane

ci ¼ ciFeed ;

B:C:2

r ¼ IDMembrane

qci ¼0 qr

vr ðrÞ ¼

Vr ODMembrane r 2

ð13:20Þ

ð13:21Þ

An overall mass balance of the whole system has to be coupled to previous equations for comparison with measurements. Species-concentration proﬁles, obtained by integrating Equation (13.20), are plotted in Figure 13.7 for the case of two reactions in series (A ! B ! C) occurring in a catalytic hollow-ﬁber membrane. The reactant A contained in the bulk phase on shell-side ﬂow through the membrane where reacting produces the intermediate product B; then, B is converted in the ﬁnal product C. Variations on concentration proﬁles are present only inside the ﬁber, outside the ﬁber there is no variation due to the reaction. No diffusion limitation in the ﬁlms were considered in the present model in order to focus on transformation inside the membrane.

13.5 Thermodynamic Equilibrium in Pd-Alloy Membrane Reactor

The product removal from the reaction volume drives the conversion that can exceed that imposed by thermodynamics to the TR (TREC, TR equilibrium conversion). An MR has to respect the thermodynamic law, even if it exceeds the TREC. Therefore, an upper limit to conversion of an MR has to be identiﬁed (MREC, MR equilibrium conversion). The permeation equilibrium has to be reached in an MR in addition to the reaction equilibrium typical of a TR. This means no

j301

j 13 Fundamental of Chemical Membrane Reactors

302

further permeation – no net permeating ﬂux through the membrane – and it can be expressed: Permeating Ji ¼ 0 , f PiReaction ; PiPermeation ¼ 0 , PiReaction ¼ PiPermeation 8 permeable specie i This relation has to be coupled to the chemical equilibrium law Kpj ðTÞ ¼

NY species

n

Pi i;j

8 reaction j

ð13:22Þ ð13:23Þ

i

Since the permeance and permeability are always different from zero, no permeation is equivalent to zero permeation driving force, which occurs when the species partial pressures on both membrane sides are equal to each other. It must be noted that the equilibrium conversion of an MR is independent of the permeation law that expresses the penetrant velocity through the membrane materials. The MREC is a function of the thermodynamic variables (i.e., temperature and pressure) and initial compositions on both sides of the Pd-alloy membranes. MREC ¼ MR equilibrium conversion Sweep Þ ¼ f ðKp; T Reaction ; P Reaction ; YiFeed ; T Permeation ; PPermeation ; F Feed =F Sweep ; Yi ð13:24Þ MREC, as the TREC, does not depend on the reaction path. In addition, there is no dependence on the membrane-permeation properties (related to the time required for species permeation).1) In any case, the ﬁnal value reached depends on the extractive capacity of the system, for example, the pressure and composition on the permeate side. The composition on the permeate side, similarly to the feed molar ratio, can be expressed by considering the ratio (named sweep factor) between the initial molar number of nonpermeating species (present on the permeate side) and the initial molar number of the reference reactant, for example, methane for methane steam reforming, orcarbonmonoxideforwatergas shift). Thesweepfactorwas deﬁnedfora closedMRas: Initial time nSweep ð13:25Þ I ¼ Sweep factor ¼ Feed nReference component The sweep factor can be deﬁned in analogous way for an open MR such as a plugﬂow, in this case ﬂow rates of feed and sweep streams are used instead of the number of moles: F Sweep I ¼ Sweep factor ¼ Feed ð13:26Þ FReference component The MREC was evaluated for some dehydrogenation reactions (e.g., methane steam reforming and water gas shift) in a Pd-based MR where the membranes are characterized by inﬁnite selectivity towards hydrogen [36, 37]. The signiﬁcant 1) A very long time, theoretically inﬁnite, is required by a closed TR and MR to reach equilibrium conversion.

13.6 Conclusions

importance of the permeation on conversion is shown in Figure 13.9 reporting the MREC in Pd-based MRs for methane steam reforming and water gas shift reactions. MREC functionality with the temperature is the same as that of a TR, but the pressure always has a positive effect, since it drives the permeation. In particular, for methane steam reforming the feed pressure has an opposite effect to that shown by a TR.

13.6 Conclusions

Fundamental aspects of chemical membrane reactors (MRs) were introduced and discussed focusing on the peculiarity of MRs. Removal by membrane permeation is the novel term in the mass balance of these reactors. The permeation through the membrane is responsible for the improved performance of an MR; in fact, higher (net) reaction rates, residence times, and hence improved conversions and selectivity versus the desired product are realized in these advanced systems. The permeation depends on the membranes and the related separation mechanism; thus, some transport mechanisms were recalled in their principal aspects and no deep analysis of these mechanisms was proposed. Owing to the permeation, the energy transport in MRs requires further consideration on the conﬁguration of an MR to be used, speciﬁcally, in energy-intensive reactions for example, methane steam reforming. The energy transport drives to the right MR conﬁguration to be used: the catalyst inside the membrane core or in the annular space. The two volumes have very different heat transfer from the energy source and reaction volume. In addition, a higher conversion of an MR also means a higher energy demand. Therefore, the energy management in an MR is also more important than that in a TR; the temperature being a fundamental variable for the reaction and permeation rate. The permeation effect is also shown by the suited reduction of the reaction volume (or catalyst amount) (Figure 13.11); in fact, the MR reaction volume is signiﬁcantly lower than that of a TR.

Figure 13.11 Reaction volume (catalyst amount) of an MR with respect to a TR for water gas shift reaction.

j303

j 13 Fundamental of Chemical Membrane Reactors

304

List of Symbols

A C Cp d D Driving force Sievert DPH2 E F h I ID J Keq Kp kz L n N OD P Pe0 Permeability Permeance Permeating ﬂux R r rij T t U V VI X Yi j Y z

Surface area (m2) Concentration (mol m3) Speciﬁc heat (J mol1 K1) Diameter (m) Diffusivity (m2 s1) Pa (Pa0.5) H2 permeation Sieverts driving force (Pa0.5) Activation energy (J mol1) Molar ﬂow rate, (mol s1) Enthalpy (J mol1) Sweep factor () Inner diameter (m) Permeating ﬂux (mol m2 s1) Equilibrium constant () Equilibrium constant in terms of partial pressures () Axial thermal conductivity (J m1 s1 K1) Length (m) Number of mole () Molar ﬂux (mol m2 s1) Outer diameter (m) Pressure (Pa) Permeability pre-exponential factor (mol m1 s1 Pa0.5) mol m1 s1 Pa0.5 mol m2 s1 Pa0.5 mol m2 s1 Gas law constant (82.05 cm3 atm g-mol1 K1) Radial coordinate (m) jth reaction rate for ith species (mol m3 s1) Temperature ( C or K) Time (s) Overall heat-transfer coefﬁcient (W m2 K1) Volume (m3) Volume index () Conversion () Molar fraction of the species ith () Molar fraction vector of the jth stream () Axial coordinate (m)

Greek Letters

F P

Enthalpy ﬂux associated to hydrogen permeation (J m2 s1) Permeance (mol m2 s1 Pa0.5)

13.6 Conclusions

p Y d e nI,j r t

Pressure matrix Heat generated by chemical reactions, J m3 s1 Membrane thickness (m) Porosity () Stoichiometric coefﬁcient with respect to the reference component of ith species in jth reaction () Density (g m3) Space time, s and tortuosity ()

Superscripts

Annulus Feed Lumen Membrane Permeate Permeating Permeation, permeation side Reaction, Reaction side Shell Sweep

Annulus side in a luminal (tubular) MR Membrane module inlet stream referred to Lumen side in a tubular MR Membrane phase referred to Membrane module permeate stream referred to Membrane module permeating stream referred to Membrane module stream on the permeation volume referred to Membrane module stream on the reaction volume referred to Membrane module shell side referred to Membrane module inlet stream on permeate side referred to

Acronyms

B.C. I.C. MR MREC MSR PDE TR TREC WGS

Boundary condition Initial condition Membrane reactor Membrane reactor equilibrium conversion Methane steam reforming Partial differential equation Traditional reactor Traditional reactor equilibrium conversion Water gas shift

Acknowledgments

The Italian Ministry of Education, University and Research, Progetto FIRB–CAMERE RBNE03JCR5 – Nuove membrane catalitiche e reattori catalitici a membrana per

j305

j 13 Fundamental of Chemical Membrane Reactors

306

reazioni selettive come sistemi avanzati per uno sviluppo sostenibile is gratefully acknowledged for cofunding this work.

References 1 Brunetti, A., Caravella, C., Barbieri, G. and Drioli, E. (2007) Simulation study of water gas shift in a membrane reactor. Journal of Membrane Science, 306 (1–2), 329–340. 2 Barbieri, G., Brunetti, A., Tricoli, G. and Drioli, E. (2008) An innovative conﬁguration of a Pd-based membrane reactor for the production of pure hydrogen. Experimental analysis of water gas shift. Journal of Power Sources, 182 (1), 160–167, http://dx.doi.org/10.1016/ j.jpowsour.2008.03.086. 3 Stankiewicz, A. and Moulijn, J.A. (2002) Process intensiﬁcation. Industrial & Engineering Chemistry Research, 41, 1920. 4 Drioli, E. and Romano, M. (2001) Progress and new perspectives on integrated membrane operations for sustainable industrial growth. Industrial & Engineering Chemistry Research, 40, 1277–1300. 5 Brunetti, A., Barbieri, G. and Drioli, E. (2008) A PEM-FC and H2 membrane puriﬁcation integrated plant. Chemical Engineering and Processing: Process Intensiﬁcation, 47 (7), 1081–1089, http:// ds.doi.org/10.1016/j.cep.2007.03.015, special issue Euromembrane 2006. 6 Kikuchi, E., Uemiya, S., Sato, N., Inoue, H., Ando, H. and Matsuda, T. (1989) Membrane reactor using microporous glass-supported thin ﬁlm of palladium. Application to the water–gas shift reaction. Chemistry Letters, 489. 7 Uemiya, S., Sato, N., Inoue, H., Ando, H. and Kikuchi, E. (1991) The water–gas shift reaction assisted by palladium membrane reactor. Industrial & Engineering Chemistry Research, 30, 585. 8 Criscuoli, A., Basile, A. and Drioli, E. (2000) An analysis of the performance of membrane reactors for the water–gas shift

9

10

11

12

13

14

15

16

reaction using gas feed mixtures. Catalysis Today, 56, 53. Barbieri, G. and Bernardo, P. (2004) Experimental evaluation of hydrogen production by membrane reaction, in Carbon Dioxide Capture for Storage in Deep Geologic Formations – Results from the CO2 Capture Project, vol. 1, Elsevier, pp. 385–408, Chapter 22. Onstot, W.J., Minet, R.G. and Tsotsis, T.T. (2001) Design aspects of membrane reactors for dry reforming of methane for the production of hydrogen. Industrial & Engineering Chemistry Research, 40, 242–251. Sjardin, M., Damen, K.J. and Faaij, A.P.C. (2006) Techno-economic prospects of small-scale membrane reactors in a future hydrogen-fuelled transportation sector. Energy, 31 (14), 2187–2219. Bracht, M., Alderliesten, P.T., Kloster, R., Pruschek, R., Haupt, G., Xue, E., Ross, J.R.H., Koukou, M.K. and Papayannakos, N. (1997) Water gas shift membrane reactor for CO2 control in IGCC systems: Techno-economic feasibility study. Energy Conversion and Management, 38 (9999), S159–S164. Yeung, K.L., Zhang, X., Lau, W.N. and Martin-Aranda, R. (2005) Experiments and modeling of membrane microreactors. Catalysis Today, 110 (1–2), 26–37. Dudukovic, M.P. (1999) Trends in catalytic reaction engineering. Catalysis Today, 48 (1–4), 5–15. Sun, J., Desjardins, J., Buglass, J. and Liu, K. (2005) Noble metal water gas shift catalysis: Kinetics study and reactor design. International Journal of Hydrogen Energy, 30 (11), 1259–1264. Majumder, D. and Broadbelt, L.J. (2006) A multiscale scheme for modeling catalytic

References

17

18

19

20

21 22

23

24

25

26

27

ﬂow reactors. AIChE Journal, 52 (12), 4214–4228. Collins, J.P., Way, J.D. and Kraisuwansarn, N. (1993) A mathematical model of a catalytic membrane reactor for the decomposition of NH3. Journal of Membrane Science, 77 (2–3), 265–282. Chmielewski, D., Ziaka, Z. and Manousiouthakis, V. (1999) Conversion targets for plug ﬂow membrane reactors. Chemical Engineering Science, 54, 2979–2984. Belgued, M., Pareja, P., Amariglio, A. and Amariglio, H. (1991) Conversion of methane into higher hydrocarbons on platinum. Nature, 352, 789–790. Koerst, T., Deelen, M.J.G. and Van Santen, R.A. (1992) Hydrocarbon formation from methane by a low-temperature two-step reaction sequence. Journal of Catalysis, 138, 101. Gryaznov, V. (1999) Membrane catalysis. Catalysis Today, 51, 391–395. Armor, J.N. (1998) Applications of catalytic inorganic membrane reactors to reﬁnery products. Journal of Membrane Science, 147, 217–233. Coronas, J. and Santamaria, J. (1999) Catalytic reactors based on porous ceramic membranes. Catalysis Today, 51, 377–389. Sirkar, K.K., Shanbhag, P.V. and Kovvali, A.S. (1999) Membrane in a reactor: a functional perspective. Industrial & Engineering Chemistry Research, 38, 3715–3727. Drioli, E., Basile, A. and Criscuoli, A. (2000) High temperature membrane reactors for clean productions. Clean Products and Processes, 2, 179–186. Julbe, A., Farrusseng, D. and Guizard, C. (2001) Porous ceramic membranes for catalytic reactors - overview and new ideas. Journal of Membrane Science, 181, 3–20. Tosti, S., Bettinali, L. and Violante, V. (2000) Rolled thin Pd and Pd–Ag membranes for hydrogen separation and production. International Journal of Hydrogen Energy, 25, 319–325.

28 Belgued, M., Amariglio, A., Pareja, P. and Amariglio, H. (1996) Oxygen-free conversion of methane to higher alkanes through an isothermal two-step reaction on ruthenium. Journal of Catalysis, 161, 282–291. 29 She, Y., Dardas, Z., Gummalla, M., Vanderspurt, T. and Emerson, S. (2005) Integrated water gas shift (WGS) Pd membrane reactors for compact hydrogen production systems from reforming of fossil fuels. ACS Division of Fuel Chemistry, Preprints, 50 (2), 561. 30 Xu, Z.-Q., Chen, Q.-L. and Lu, G.-Z. (1999) The use of membrane reactor in the reaction of ethylbenzene dehydrogenation to styrene. Petrochemical Technology, 28 (6), 362. 31 Xu, Z., Chen, Q. and Lu, G. (1999) Use of membrane reactor in reaction of ethylbenzene dehydrogenation to styrene. Shiyou Huagong/Petrochemical Technology, 28 (6), 358–362. 32 Assaf, E.M., Jesus, C.D.F. and Assaf, J.M. (1998) Mathematical modelling of methane steam reforming in a membrane reactor: An isothermic model. Brazilian Journal of Chemical Engineering, 15 (2), 160–166. 33 Lin, Y.-M., Liu, S.-L., Chuang, C.-H. and Chu, Y.-T. (2003) Effect of incipient removal of hydrogen through palladium membrane on the conversion of methane steam reforming: Experimental and modelling. Catalysis Today, 82 (1–4), 127–139. 34 Capobianco, L., Del Prete, Z., Schiavetti, P. and Violante, V. (2006) Theoretical analysis of a pure hydrogen production separation plant for fuel cells dynamical applications. International Journal of Hydrogen Energy, 31, 1079–1090. 35 Barbieri, G., Scura, F. and Brunetti, A. (2008) Mathematical modelling of Pd-alloy membrane reactors, in Inorganic Membranes: Synthesis, Characterization and Applications, vol. 13 (eds R. Mallada and M. Menendez), Membrane Science and Technology, Elsevier BV, Chapter 9 (ISSN

j307

j 13 Fundamental of Chemical Membrane Reactors

308

36

37

38

39

40

41

42

0927-5193, 10.1016/S0927-5193(07) 13009-6). Marigliano, G., Barbieri, G. and Drioli, E. (2003) Equilibrium conversion for a Pd-alloy membrane reactor. Dependence on the temperature and pressure. Chemical Engineering and Processing, 42 (3), 231–236. Barbieri, G., Scura, F. and Drioli, E. (2006) Equilibrium of a Pd-alloy membrane reactor. Desalination, 200 (1–3), 679–680. Marigliano, G., Barbieri, G. and Drioli, E. (2001) Effect of energy transport in a palladium based membrane reactor for methane steam reforming process. Catalysis Today, 67 (1–3), 85–99. Bernardo, P., Algieri, C., Barbieri, G. and Drioli, E. (2008) Hydrogen puriﬁcation from carbon monoxide by means of selective oxidation using zeolite membranes. Separation and Puriﬁcation Technology, 62, 631–637. Ward, T.L. and Dao, T. (1999) Model of hydrogen permeation behavior in palladium membranes. Journal of Membrane Science, 153, 211–231. Caravella, A., Barbieri, G. and Drioli, E. (2008) Modelling and simulation of hydrogen permeation through supported Pd-alloy membranes with a multicomponent approach. Chemical Engineering Science, 63 (8), 2149–2160, 10.1016/j.ces.2008.01.009. Barbieri, G., Scura, F., Lentini, F., De Luca, G. and Drioli, E. (2008) A novel model equation for the permeation of hydrogen

43

44

45

46

47

48

in mixture with carbon monoxide through Pd-Ag membranes. Separation and Puriﬁcation Technology, 61 (2), 217–224, 10.1016/j.seppur.2007.10.010. Xu, S.J. and Thomson, W.J. (1999) Oxygen permeation rates through ion-conducting perovskite membranes. Chemical Engineering Science, 54, 3839–3850. Wang, h., Tablet, C., Schiestel, T., Werth, S. and Caro, J. (2006) Partial oxidation of methane to syngas in perovskite hollow ﬁber membrane reactor. Catalysis Communications, 7, 907–912. Buonomennna, M.G., Lopea, L.C., Barbieri, G., Favia, P., dAgostino, R. and Drioli, E. (2007) Sodium tungstate immobilized on plasma-treated PVDF membranes: new efﬁcient heterogeneous catalyst for oxidation of secondary amines to nitrones. Journal of Molecular Catalysis. A, Chemical, 273, 32–38, 10.1016/ j.molcata.2007.03.065. Bonchio, M., Carraro, M., Gardan, M., Scorrano, G., Drioli, E. and Fontananova, E. (2006) Hybrid photocatalytic membranes embedding decatungstate for heterogeneous photooxygenation. Topics in Catalysis, 40 (1–4), 133–140. Caravella, A., Barbieri, G. and Drioli, E., (2009) Concentration polarization analysis in self-supported Pd-based membranes. Separation and Puriﬁcation Technology, in press, http://dx.doi.org/10.1016/ j.seppur.2009.01.008. Bird, R.B., Stewart, W.E. and Lightfoot, E.N. (1979) Fenomeni di Trasporto, Casa Editrice Ambrosiana, Milano.

j309

14 Mathematical Modeling of Biochemical Membrane Reactors Endre Nagy

14.1 Introduction

Membrane bioreactor (MBR) technology is advancing rapidly around the world both in research and commercial applications [1–4]. Integrating the properties of membranes with biological catalyst such as cells or enzymes forms the basis of an important new technology called membrane bioreactors. The membrane layer is especially useful for immobilizing whole cells (bacteria, yeast, mammalian, and plant cells) [5, 6], bioactive molecules such as enzymes [7–9] to produce a wide variety of chemicals and substances. The MBR were introduced over 30 years ago and until now they are recommended or applied for production of foods, biofuels, plant metabolites, amino acids, antibiotics, anti-inﬂammatories, anticancer drugs, vitamins, proteins, optically pure enantiomers, isomers, ﬁne chemicals, as well as for treatment of wastewater (e.g., industrial, domestic, and municipal [2, 10]). Our work is focused primarily on the hollow-ﬁber bioreactor with biocatalyst, either live cells or enzymes, inoculated into the shell or immobilized within the membrane matrix or in a thin layer at the membrane matrix-shell interface. Membrane bioreactors for immobilized whole cells can provide a suitable environment for high cell densities [8, 11, 12]. Cells are either grown in the extracapillary space with medium ﬂow through the ﬁbers and supplied with oxygen and nutrients, or grown within the ﬁbers with medium ﬂow outside or across the ﬁbers, while wastes and desired products are removed. The main advantages of the hollow-ﬁber bioreactor are the large speciﬁc surface area (internal and external surface of the membrane) for cell adhesion or enzyme immobilization; the ability to grow cells to high density; the possibility for simultaneous reaction and separation; relatively short diffusion path inthemembrane layer; thepresence of convective velocity through the membrane if necessary in order to avoid the nutrient limitation [13, 14]. The performance of a hollow-ﬁber or sheet bioreactor is primarily determined by the momentum and mass-transport rate [15, 16] of the key nutrients through the biocatalytic membrane layer. Thus, the operating conditions (transmembrane pressure, feed velocity), the physical properties of membrane (porosity, wall thickness, lumen radius, matrix structure, etc.) can considerably inﬂuence the performance of a bioreactor, the

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 14 Mathematical Modeling of Biochemical Membrane Reactors

310

effectiveness of the reaction. The main technological difﬁculties in using membrane bioreactors on an industrial level are related to rate-limiting aspects and scale-up difﬁculties of this technology. The limited transport of nutrients can cause serious damage in production [12, 14]. The introduction of convective transport is crucial in overcoming diffusive mass-transport limitation of nutrients [17] especially of the sparingly soluble oxygen. The mathematical description of the transport processes enables us to predict the concentration distributions of nutrients in the catalyst membrane layer, and thus, it makes it possible to choose the correct operating conditions that provide a sufﬁcient level of nutrient concentration in the membrane layer. The main aim of this study to give closed, as simple as possible, mathematical equations in order to predict the concentration distribution and the mass-transfer rate through a biocatalyst membrane layer as well as the concentration variation in the lumen (or shell) side of a capillary membrane with particular regard to the variable transport parameters (diffusion coefﬁcient, convective velocity, reaction rate constant) due to the anisotropy of the membrane and/or cell colony in or around the membrane [13, 14].

14.2 Membrane Bioreactors with Membrane as Bioreactor

Membrane bioreactors have been reviewed previously in every detail [3, 4, 7, 8, 18]. There are two main types of membrane bioreactors: (i) the system consists of a traditional stirred-tank reactor combined with a membrane separation unit (Figure 14.1); (ii) the membrane contains the immobilized biocatalysts such as enzymes, micro-organisms and antibodies and thus, acts as a support and a separation unit (Figure 14.2). The biocatalyst can be immobilized in or on the membrane by entrapment, geliﬁcation, physical adsorption, ionic binding, covalent binding or crosslinking [3, 7, 18]. Our attention will be primarily focused on the second case where the membrane acts as a support for biocatalyst and as a separation unit, in this study. The momentum and mass-transport process, in principle, are the same in both cases, namely when there is

Figure 14.1 Schematic illustration of the external membrane bioreactor configuration.

14.2 Membrane Bioreactors with Membrane as Bioreactor

Figure 14.2 Membrane bioreactor with immobilized biocatalysts (enzyme or micro-organism).

no biochemical reaction in the membrane (Figure 14.1) or when a biochemical reaction occurs in the membrane due to its catalytic activity (Figure 14.2). 14.2.1 Enzyme Membrane Reactor

The enzymatic bioconversion processes are of increasing use in the production, transformation and valorization of raw materials [7]. Important applications have been developed in the ﬁeld of food industries, ﬁne chemicals or even for environmental purposes. Several important applications of the enzyme membrane reactor are given in Charossets [8] and Rioss papers [7] as well as an excellent summary given by Giorno and Drioli [3] and Miguel et al. [9]. They discussed the different enzyme–membrane conﬁgurations (recycle-, dialysis-, diffusion-, multiphase membrane reactors) and the different types of enzyme retention (trapping, chemical coupling, adsorption electrostatic interactions, etc.). The enzyme membrane reactor has several advantages (continuous mode, retention and reuse of catalyst, reduction in substrate/product inhibition, enzyme-free product, integrated process, etc.) and also some disadvantages (decreasing enzyme activity as a function of time, membrane fouling, low substrate concentration, etc.). The enzyme can be immobilized into the membrane matrix or on the membrane interface. In this latter case, the membrane surface is covered by a gel layer and the enzyme binds to this layer [19, 20]. In this case the substrate solution has direct contact with the catalyst, while in the other case, namely when the enzyme is placed in the membrane matrix, the substrate has to ﬂow through the membrane layer to come into contact with the biocatalyst. Concerning the mathematical description of these enzymatic processes, basically two different cases can be distinguished, namely the enzyme or live cells are immobilized onto or in the membrane layer or the biocatalyst is dissolved homogeneously or mixed heterogeneously (it is immobilized in porous particles) in the feed phase. The

j311

j 14 Mathematical Modeling of Biochemical Membrane Reactors

312

description of these two biochemical processes is different (discussed later). The structure of the membrane matrix does not practically change during the biocatalytic process, thus, the transport parameters remain constant. Obviously, due to the fouling or cake forming, if it is the case, the increase of the external mass-transfer resistance can alter parameters such as convective velocity, external mass-transfer coefﬁcient. This change has to be taken into account to describe the transport process. 14.2.2 Whole-Cell Membrane Bioreactor

Membrane bioreactors for immobilized whole cells [6, 12, 21] provide an advantages environment for increased cell densities. The cells are perfused via a membrane with a steady continuous ﬂow of medium containing the oxygen and other nutrients. The cells are either grown in the extracapillary space (to form a bioﬁlm), or grown within the ﬁbers (Figure 14.2). It was shown that a mass-transfer limitation for oxygen or other nutrient could occur, especially at higher cell density [5, 14, 21]. Due to the change of the nutrient concentration in the axial direction, the density of the cell culture, the thickness of the bioﬁlm on the membrane interface can also change. This fact can alter the values of transport parameters (diffusion coefﬁcient, convective velocity, it can even alter the biochemical reaction rate (the consumption rate of nutrient). Theoretical studies alsoconﬁrm that nutrient limitation can often occur in hollow-ﬁber biocatalytic membranes [10, 13, 16]. To avoid this limitation the mass-transfer rate through the cell culture should be increased. This can be achieved by a construction change, by a change of the membrane structure, thickness or for example, by the increase of the transmembrane pressure. Due to this, the radial convective velocity also increases. Applications of whole-cell biocatalytic membrane reactors, in the agro-food industry and in pharmaceutical and biomedical treatments are listed by Giorno and Drioli [3]. Frazeres and Cabral [9] have reviewed the most important applications of enzyme membrane reactors such as hydrolysis of macromolecules, biotransformation of lipids, reactions with cofactors, synthesis of peptides, optical resolution of amino acids. Another widespread application of the membrane bioreactor is the wastewater treatment will be discussed in a separate section.

14.3 Membrane Bioreactors with Membrane as Separation Unit 14.3.1 Moving-Bed Biofilm Membrane reactor

The basic advantage of the suspended bioﬁlm membrane reactors over the suspended biomass system (either with dispersed cells or with ﬂocs) is that the former are able to retain much more biomass [22]. It substantially reduces the biomass wash out and allows a more stable operation with higher biomass concentration. A moving-bedbioﬁlm membrane reactor is illustrated in Figure 14.5. The biomass is immobilized inside and outside of the ﬂuidized particles. The structure of the bioﬁlm continuously

14.3 Membrane Bioreactors with Membrane as Separation Unit

changes during the growing period of the micro-organism. In this study we do not analyze in detail the mass transfer into the suspended particles. The general equations are known and they are excellently summarized by Melo and Oliviera [22]. 14.3.2 Wastewater Treatment by Whole-Cell Membrane Reactor

In this case, a biological step and a membrane module are integrated where both of them have speciﬁc functions [22]: (i) biological degradation of organic pollutant is carried out in a traditional bioreactor by micro-organisms; (ii) separation of micro-organisms from the treated wastewater is performed by the membrane module. The membrane enables recycling of the activated sludge to the bioreactor as well as the production of cleaned water. There are here also two types of conﬁgurations for the membrane array in the wastewater treatment: the membrane can be placed outside (Figure 14.1) or inside of the bioreactor (Figure 14.3). For the external conﬁguration, the treated liquid is ﬁltered under pressure in a speciﬁc membrane module, whereas for the submerged conﬁguration, the ﬁltration is carried out under vacuum. The latter conﬁguration seems to be more economical based on energy consumption: a recycle pump is not needed since the aeration generates a tangential ﬂow around the submerged membrane ﬁbers [22]. The cells here are not immobilized in/or on the membrane layer, but bioﬁlm can be formed on the membrane interface causing mass-transport difﬁculties through the membrane. The transport through the membrane is simple ﬁltration that can be altered by fouling and/or the bioﬁlm formed during the treatment. The wastewater treatment is widely applied for industrial (e.g., food [23], beverage [24], dairy industry [25], municipal [26, 27]) as well as domestic wastewater. 14.3.3 Membrane Fouling

As the liquid passes through the membrane in crossﬂow ﬁltration, the particles, macromolecules, colloids, and so on, rejected by the membrane will accumulate in

Figure 14.3 Schematic illustration of immersed membrane bioreactor configurations.

j313

j 14 Mathematical Modeling of Biochemical Membrane Reactors

314

Figure 14.4 Schematic of the membrane, cake (gel) and concentration-polarization layer with their parameters.

the immediate vicinity of the membrane surface to form a dense layer (cake layer) of retained particles (Figure 14.4). The fouling components of pore clogging, deposition of large particles, sludge cake growth during wastewater treatment are dynamic processes, it induces transmembrane ﬂux reduction. When the permeation rate reaches a critical value, membrane washing becomes necessary. Many studies have focused on the above problems [1, 28–31]. It is shown that the forming of a cake layer, gel layer or bioﬁlm depends strongly on the permeate ﬂux as well as on the membrane structure. Conventional techniques for limiting membrane fouling: its reduction by aeration in the vicinity of the membrane, by ﬁltration below the critical ﬂux, by the addition of coagulants, by high-frequency backpulsing, or by utilizing a high recycle velocity and/or removal of the fouling material by chemical washing (backwashing or backpulsing). Figure 14.3 also shows the concentration-polarization layer that also forms during the ﬁltration. Due to it, there is a backdiffusion of the retained compound that has higher concentration on the membrane surface. These phenomena can also decrease the efﬁciency of the ﬁltration. This effect should also be taken into account during the mathematical modeling of the transport processes of the membrane bioreactor, as will be discussed later.

14.4 Mathematical Modeling of Membrane Bioreactor 14.4.1 Modeling of Enzyme Membrane Layer/Biofilm Reactor

The principle of the mass transport of substrates/nutrients into the immobilized enzyme/cells, through a solid, porous layer (membrane, bioﬁlm) or through a gel layer of enzyme/cells is the same. The structure, the thickness of this mass-transport layer can be very different, thus, the mass-transport parameters, namely diffusion

14.4 Mathematical Modeling of Membrane Bioreactor

coefﬁcient, convective velocity, the bioreaction rate constant, their dependency on the concentration and/or space coordinate is characteristic of the porous layer and the nature of the biocatalysts. Several investigators modeled the mass transport through this biocatalyst layer, through enzyme membrane layer [32–36] or cell-culture membrane layer [5, 12–14, 22, 37–39]. Some assumptions made for expression of the differential mass-balance equation to the biocatalytic membrane layer: . . . . . .

Reaction occurs at every position within the biocatalyst layer; Reaction has one rate-limiting substrate/nutrient; Mass transport through the biocatalyst layer occurs by diffusion and convection; The partitioning of the components (substrate, product) is negligible; The mass-transport parameters (diffusion coefﬁcient, convective velocity, bioreaction rate constant) can vary as a function of the space coordinate; The external mass-transfer resistance is to be taken into account.

Thus, the mass-balance equation obtained for various geometries, perpendicular to the membrane interface, can be given as follows: 1 q qc ðp þ 1Þ qðADcÞ qðAncÞ qc DA þ Q ¼ ð14:1Þ A qr qr r qr dr qt where A is the real area for mass transport, c is the substrate concentration, D is the diffusion coefﬁcient, r is the radial space coordinate, t is the time, Q is the reaction rate, p is a geometrical factor with values of 1 for spherical pellets, 0 for cylindrical coordinates and 1 for rectangular membranes [32]. The value of A can be changed with the porosity of the biocatalyst membrane layer. The variation of the masstransport parameters can especially occur during the growth of cells around/in hollow ﬁbers [12, 14] because of their inhomogeneous growth due to the variation of the nutrient concentration. Close to the entrance of the bioreactor the density of nutrients (and the thickness of the bioﬁlm formed on the membrane surface) could be much higher due to the higher nutrient concentration than that at the end of the reactor where the nutrient concentration can be much lower. The density of the cell can change not only the values of transport parameters but also the value of the reaction rate constant [12]. The variation of cell density is also true in the biocatalyst membrane layer perpendicular to the inlet surface. Increasing distance from the surface can mean decreasing nutrient concentration. This is why the variability of the transport parameters should also be taken into account. The source term can be different in biocatalytic reactions, the most often applied equations are listed in Table 14.1. The inhibited reaction can take place in both the enzymatic and microbial reactions. For the sake of simpliﬁcation, let us regard a steady-state reaction with constant area for mass transport (A ¼ constant) and let us use the Cartesian coordinate ( p ¼ 1), in the following. Thus, Equation 14.2 can be obtained by rewriting of Equation 14.1, as follows (y is here the transverse space coordinate, perpendicular to the membrane interface): d dc dðucÞ D Q ¼ 0 ð14:2Þ dy dy dy

j315

j 14 Mathematical Modeling of Biochemical Membrane Reactors

316

Table 14.1 Expressions of the important biocatalytic reactions.

rmax c KMi þ c rmax c Substrate inhibition: Q ¼ KMi þ c þ c 2 =Kc Michaelis–Menten kinetics: Q ¼

Substrate inhibition and competitive product inhibition: Q ¼

rmax c KMi ð1 þ p=Kp Þ þ c þ c 2 =Kc

rmax c KMi ð1 þ p=Kp Þ þ c rmax c Noncompetitive product inhibition: Q ¼ ðKMi þ cÞð1 þ p=Kp Þ Competitive product inhibition: Q ¼

In general case, as was mentioned, the diffusion coefﬁcient and/or convective velocity can depend on the space coordinate, thus D ¼ D(y), u(y), [or on the concentration, D ¼ D(c) or both of them, D ¼ D(c, y)]. In the boundary conditions the external mass-transfer resistance is also taken into account. Thus, the boundary conditions will be as follows: dc1 o þ u1 c1 jy¼0 þ if x ¼ 0 then uin c1L þ b0 ðc1 c1L Þ ¼ D1 dx y¼0 þ ð14:3aÞ if

x¼d

then

uout cML

þ bd ðcML c2o Þ

dcM ¼ DM þ uM cM jy¼d dx y¼d ð14:3bÞ

c denotes the concentration of liquid at the membrane interface, c1o , c2o denote the bulk concentration of feed and downstream, respectively, d is the membrane thickness. The membrane concentration, c is given here in units of gmol/m3. This can be easily obtained by means of the usually applied in, for example, g/g unit of measure with the equation of c ¼ wr/M, where w concentration in kg/kg, r – membrane density, kg/m3, M – molar weight, kg/mol. Its dimensionless form can be given by C ¼ c=ðc1o HÞ [the H is the partition coefﬁcient, its value is mostly unity during biochemical reactions]. A solution methodology of the above, a nonlinear differential equation, will be shown. In essence, this solution method serves the mass-transfer rate and the concentration distribution in closed, explicit mathematical expression. The method can be applied for Cartesian coordinates and cylindrical coordinates, as will be shown. For the solution of Equation 14.2, the biocatalytic membrane should be divided into M sublayers, in the direction of the mass transport, that is perpendicular to the membrane interface (for details see e.g., Nagys paper [40]), with thickness of Dd (Dd ¼ d/M) and with constant transport parameters in every sublayer. Thus, for the mth sublayer of the membrane layer, using dimensionless quantities, it can be obtained: Dm

d 2 cm dcm nm km cm ¼ 0 dy2 dy

xm1 < x < xm :

ð14:4Þ

14.4 Mathematical Modeling of Membrane Bioreactor

The value of km can be obtained according, for example, to the Michaelis–Menten kinetics as follows: km ¼

rmax KMi þ cm

ð14:5Þ

In dimensionless form one can get the following equation: d2 Cm dCm Ha2m Cm ¼ 0 Pem ð14:6Þ dY 2 dY qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ where Y ¼ y/d; Pem ¼ umd/Dm; Ham ¼ d 2 km =Dm . The Ha number (Ham) deﬁned here for the mth sublayer of the membrane layer corresponds to the well-known Thiele modulus deﬁned for catalyst particles, while the Peclet number corresponds to the often used Bodenstein number. Introducing the following equation: PeY ~ ð14:7Þ C ¼ Cexp 2 One can get: d2 C~m Q2m C~m ¼ 0 dY 2

ð14:8Þ

with rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Pe2m þ Ha2m Qm ¼ 4

ð14:9Þ

The solution of Equation 14.6 can be easily obtained by well-known mathematical methods as follows: ~

Cm ¼ Tm eðlm YÞ þ Pm eðlm YÞ

Ym1 < Y < Ym

ð14:10Þ

with lm ¼

Pem Qm 2

~m ¼ Pem þ Qm l 2

The Tm and Pm parameters of Equation 14.10 can be determined by means of the boundary conditions for the mth sublayer (with 1 m M). It will be shown in the following how the values of Tm and Pm can be obtained. The boundary conditions at the internal interfaces of the sublayers (1 m M 1; Ym ¼ mDY; DY ¼ 1/M) can be obtained from the following two equations:

dCm Dm þ 1 dCm þ 1 þ Pem Cm ¼ þ Pem þ 1 Cm þ 1 dY Dm dY

at

Y ¼ Ym ð14:11aÞ

Hm Cm ¼ Hm þ 1 Cm þ 1

at

Y ¼ Ym

ð14:11bÞ

j317

j 14 Mathematical Modeling of Biochemical Membrane Reactors

318

The general solution of the algebraic equation system obtained by means of the internal [(14.11a) and (14.11b)] conditions with m ¼ 1,2, . . ., M and external boundary conditions [(14.3a) and (14.3b)] are given in Appendix B. The mass-transfer rate on the upstream side of the membrane can be given, for that case, as follows: Jin ¼

D1 c1o dC D1 c1o ~ 1 P1 Þ þ Pe1 C ðl1 T1 þ l ¼ d d dY Y¼0

ð14:12Þ

The outlet mass-transfer rate can be similarly given. This value should be as low as possible to avoid the loss of the substrate during the process. 14.4.2 Concentration Distribution and Mass-Transfer Rates for Real Systems

In this section the concentration distribution and the mass-transfer rate of a substrate is brieﬂy discussed and shown under real operating conditions. The axial and radial depletion of substrate, for example, oxygen, nutrient, can often be critical scalelimiting factor in a cell-culture hollow-ﬁber reactor [12–14, 33]. In order to increase the substrate concentration in the membrane bioreactor, a sufﬁcient diffusion rate and/or convective ﬂow has to be provided through the lumen, in the axial direction, and through the membrane layer, in the radial direction, of the hollow ﬁber. Typical operating conditions of a hollow-ﬁber bioreactor were applied (Table 14.2) to calculate the inlet and outlet mass-transfer rates of a substrate. From this, the effectiveness of the biocatalytic reaction as well as the sufﬁciency of the nutrient supply can be estimated. The biochemical reaction rate depends on the amount of catalyst immobilized in the membrane or on the density of cells in the membrane structure. The oxygen consumption rate of cells can be estimated to be 0.1 1012 mol/(cell h) [12–14], while the cell density may be about 2 108 cell/cm3. According to this, the consumption rate is equal to 6 109 mol/cm3 s. From that we can get for the Ha number assuming the cm value in Equation 14.5 is equal to zero: Ha ¼ 0.1 (d ¼ 100 mm, D ¼ 1 1010 m2/s, k ¼ 1.44 106 1/s) or Ha ¼ 0.6 (d ¼ 500 mm, D ¼ 1 1010 m2/s, k ¼ 1.44 106 1/s). The convective velocity through the membrane is an important means to avoid the substrate limitation (Figure 14.5). With increasing Pe number, that is, with the increase of the transmembrane pressure, the Table 14.2 Membrane module characteristics and physical

parameters applied for calculation of the mass-transfer rates into and out of a sheet membrane [12–14, 33]. Pressure difference: 20–30 kPa Pe number: 0.1–10 Ha number: 0.1–5 Diffusion coefﬁcient: 109–1010 m2/s Membrane thickness: (100–1000) · 106 m Permeation velocity: 104–106 m/s

14.4 Mathematical Modeling of Membrane Bioreactor

Figure 14.5 Typical concentration distribution in the membrane at different Pe numbers (Ha = 1, d = 100 mm, D = 4 1010 m2/s, b0 = bd ! 1).

concentration can rapidly be increased in the biocatalytic membrane layer. The model presented is suitable to predict the substrate transport in the case of a nonlinear source term, that is, in the case of Michaelis–Menten (MM) kinetics or inhibition kinetics (Table 14.2). In Figures 14.6(a) and (b), the concentration distribution and the ratio of the outlet and the inlet mass-transfer rates are plotted, respectively. The calculation was made by the classical MM theory (continuous lines), and the two limiting cases, namely ﬁrst-order (KMi c, dotted lines) and zero-order (KMi c, broken lines) kinetics, were applied. The outlet concentration of the substrate was chosen to be 0.2, which was considered as the critical concentration. Below this value the micro-organism was supposed to work insufﬁciently. Let us look at the values obtained at Ha ¼ 1 and at Pe ¼ 1 for the concentration distributions (Figure 14.6a). The three models show only slight differences, which will be higher at Ha ¼ 1.9. There is a curve that has a minimum value in the ﬁgure. In order to maintain this concentration distribution, the substrate should be fed at both sides of the membrane. The substrate then might be transported form the downstream side by diffusion. Nagy [41] has investigated the diffusional mass transport through the enzyme membrane layer when the substrate can enter the membrane at both sides. The effect of an asymmetric membrane on the mass-transfer rate in the case of a ﬁrstorder biochemical reaction has been shown. The ratio of the outlet mass-transfer rate and the inlet one is an important parameter in biochemical reactions, because an essential aim of these processes should be to reduce it as low as possible. This ratio is plotted in Figure 14.7(b) as a function of the Pe number at two values of the Ha number. This ﬁgure clearly shows the difference between the three models. The other important effect on the Jout/Jin is caused by the convective velocity. The solution method presented enables prediction of the value of the Pe number that one needs to avoid the limiting substrate/nutrient concentration. The values obtained in the previous ﬁgures were predicted by constant parameters. In reality, the speciﬁc biomass concentration (and enzyme concentration) can vary as a function of space coordinate. As was mentioned, the activity of cells can be much higher at higher substrate concentration, that is, close to the feed membrane interface that can cause a higher cell density close to the interface. This can be true in the longitudinal

j319

j 14 Mathematical Modeling of Biochemical Membrane Reactors

320

Figure 14.6 (a) Concentration vs. membrane thickness applying the Michaelis–Menten kinetics (continuous lines) and its limiting kinetics, namely first-order (. . . . . .) and zeroorder (- - - -) ones. (Pe = 1, d = 100 mm, D = 4 1010 m2/s, b0 = bd ! 1). (b) The relative values of the outlet mass-transfer rates as

a function of Pe number at two different values of Ha number (the Michaelis–Menten kinetics (continuous lines) and its limiting kinetics, dotted lines, — namely first-order (. . . . . .) and zero-order (- -) ones; Pe = 1, d = 100 mm, D = 4 1010 m2/s, b0 = bd ! 1).

direction of the hollow ﬁber, as well. This inhomogeneity of biomass properties can alter the value of diffusion coefﬁcient and also that of convective velocity. These facts can be easily taken into account by means of the presented model. Figure 14.7 presents the predicted concentrationinthree different cases,namelyforconstant Pe number (Pe ¼ 5, continuous line), for linearly increasing Pe number as a function of the y coordinate [Pe (Y) ¼ 10 (1 þ 0.9i/M), broken line] and for linearly decreasing Pe number [Pe(Y) ¼ 10 (1 0.9i/M), dotted lines]. The average value of Pe number was the same in all the three cases, namely Peave ¼ 5. As can be seen, the change in Peclet number in the direction of the diffusion path of the membrane layer can essentially alter the concentration distribution in the membrane. In the case of decreasing value of Pe number, the higher starting value of Pe can substantially improve the nutrient supply across the biocatalyst layer, even far from the inlet surface. Depending on the reaction rate, the concentration can even be higher than unity, in a wide range of the membrane layer. The increasing

14.4 Mathematical Modeling of Membrane Bioreactor

Figure 14.7 Concentration distributions with constant (Pe = 5, continuous line) and with linearly – with the space coordinate – variable Peclet number: . . . . . . decreasing Pe number Pe = 10 9i/ M; i = 1–100, M = 100; – – – increasing Pe number, Pe = 1 + 9i/M. (Ha = 2, d = 100 mm, D = 4 1010 m2/s, b0 = bd ! 1).

value of Pe, that is, the increasing value of convective velocity (or decreasing value of diffusion coefﬁcient) lowers the concentration in the membrane layer. 14.4.3 Prediction of the Convective Velocity through Membrane with Cake and Polarization Layers

Generally, the pure-water ﬂux through a membrane layer, uw is directly proportional to the applied hydrostatic pressure difference (transmembrane pressure, DP) according to Darcys law as follows: uw ¼

DP ¼ Lp DP mRm

ð14:13Þ

where m is the viscosity, Rm is the hydrodynamic resistance of the membrane, uw is the hydrodynamic permeability [42]. The value of Rm is constant. However, when solutes are added to the water the behavior observed is completely different. The ﬂux does not change linearly with the pressure difference, it tends to a limiting value as a function of DP. This maximum ﬂux is called the limiting ﬂux [42–45]. Change of permeate ﬂux as a function of the transmembrane pressure difference, measured for example, by Ognier et al. [45] can be divided into three regions, a linearly increasing range (a permeate ﬂux up to about 30 dm3/m2 h), intermediate range (permeate ﬂux > 30 dm3/m2 h) and limiting ﬂux range (here, more than about 45 dm3/m2 h). In this last regime the permeate ﬂux does not increase with increasing transmembrane pressure [43]. The pore diameter of the alumina, tubular, ultraﬁltration membrane applied was 0.05 mm, its mass-transfer resistance, Rm, was 0.4 1012 m1. This anomaly of the ﬂux curve is caused by the concentration-polarization layer and by forming a cake (or gel) layer on

j321

j 14 Mathematical Modeling of Biochemical Membrane Reactors

322

the membrane surface. In the general case, three mass-transfer resistance layers determine the permeation rate through the membrane (Figure 14.4). For membrane bioreactors two important cases can be distinguished, namely: . .

The formation of a cake layer can be neglected (e.g., biocatalytic reaction without macromolecules); in this case the concentration-polarization layer can also exist; Both the concentration-polarization layer and membrane fouling are present (ﬁltration in presence of macromolecules, proteins, cells in the liquid phase).

Generally, applying the resistance-in-series model, the following expression can be given [45]: u¼

DP mðRm þ Rg þ RÞ

ð14:14Þ

where Rg is the mass-transfer resistance of the gel (cake) layer, R is the resistance of the concentration-polarization layer. For the prediction of the convective velocity through the membrane, the values of Rm, Rg, R should be determined. The ﬁltration velocity through a porous medium can be obtained for example, by the well-known Carman–Kozeny equation [46]. Thus, this equation can often be applied for both the membrane layer and the cake layer. Thus, it can be given [47]: Rg ¼

xkA2 ð1eÞ2 e3

ð14:15Þ

where x is the layer thickness, k is the Kozeny constant, A is speciﬁc surface of the cake layer, e is the porosity. Regarding only the mass-transfer rate through the membrane, the convection-diffusion model is widely used to describe the stationary permeate ﬂux during ﬁltration of small-size solutes at low concentration [43, 48–51]: d dc dðucÞ D ¼0 ð14:16Þ dy dy dy The solution of Equation 14.16 with variable parameters can be obtained by application of the method used for biochemical reaction in the previous subsection. The general solution of Equation 14.16 is as follows, in the case of constant parameters: c ¼ TexpðPeYÞ þ S

ð14:17Þ

where Y is a dimensionless coordinate, T and S are integration constants, the Peclet number, for example, for the membrane layer: Pe ¼ ud/D. The integration constants can be determined by proper boundary conditions. Let us look at the solution for the concentration-polarization layer (Figure 14.4). The boundary conditions can be given as: y ¼ 0 then c ¼ cb and y ¼ y the c ¼ c . Thus, we get (Pe ¼ uy/D): ccb expðPeYÞ1 ¼ expðPeÞ1 c cb

ð14:18Þ

14.4 Mathematical Modeling of Membrane Bioreactor

Introduction of the intrinsic retention coefﬁcient, ~r ¼ 1cp =cb , the value of the concentration modulus, c /cb can be given as: c expðPeÞ ¼ cb ~r þ ð1~r ÞexpðPeÞ

ð14:19Þ

The value of the concentration modulus depends on the convective velocity and the mass-transfer coefﬁcient of the concentration boundary layer (D/y) that means that on the membrane structure and the hydrodynamic conditions. If the retention coefﬁcient is equal to 1, then c /cb ¼ exp(Pe). The larger convective velocity (or smaller diffusion coefﬁcient) causes higher concentration polarization on the membrane interface. If one wants to take into account all three mass-transfer resistances, the massbalance equation [Equation 14.16] should be given for every layer. From that one can obtain the solution for every layer (i ¼ m for membrane layer, i ¼ g for gel layer, the concentration-polarization layer has no index according to Figure 14.4): c ¼ Ti expðPei YÞ þ Si

ð14:20Þ

The boundary conditions can be given as follows [40, 49]: ui Ci Di

dCi dCi þ 1 ¼ ui þ 1 Ci þ 1 Di þ 1 dy dy

Hi Ci ¼ Hi þ 1 Ci þ 1

ð14:21Þ ð14:22Þ

Equation 14.22 takes into account that the membrane and cake layers could have different partition coefﬁcient. Thus, you can get an algebraic equation system with 6 equations, which can be solved relatively easily (as will be discussed elsewhere). Obviously, for prediction of the ﬁltration efﬁciency, the values of transport parameters have to be known.

14.4.4 Convective Flow Profile in a Hollow-Fiber Membrane 14.4.4.1 Without Cake and Polarization Layers The basic hydrodynamic equations are the Navier–Stokes equations [51]. These equations are listed in their general form in Appendix C. The combination of these equations, for example, with Darcys law, the ﬂuid ﬂow in crossﬂow ﬁltration in tubular or capillary membranes can be described [52]. In most cases of enzyme or microbial membrane reactors where enzymes are immobilized within the membrane matrix or in a thin layer at the matrix/shell interface or the live cells are inoculated into the shell, a cake layer is not formed on the membrane surface. The concentrationpolarization layer can exist but this layer does not alter the value of the convective velocity. Several studies have modeled the convective-ﬂow proﬁles in a hollow-ﬁber and/or ﬂat-sheet membranes [11, 35, 44, 53–56]. Bruining [44] gives a general description of ﬂows and pressures for enzyme membrane reactor. Three main modes

j323

j 14 Mathematical Modeling of Biochemical Membrane Reactors

324

of operations can be classiﬁed in this case: continuous open-shell mode; closed-shell mode; suction of permeate. Considering the velocity according to Equation 14.13, and applying the Hagen–Poiseuilles law for the longitudinal pressure drop, he deﬁnes a closed mathematical expression for the pressure drop in the longitudinal direction of the ﬁber. Knowing this pressure drop in the axial direction, the pressure drop through the membrane and the convective velocity can be predicted. Song [30, 43] used the ﬁltration number to predict the pressure drop through the membrane. The simpliﬁed continuity and momentum equations for steady-state ﬂow in the ﬁber lumen and shell can be given in dimensionless form as follows [11]: qu 1 qðRuÞ þ ¼0 qX R qR

ð14:23Þ

1 q qu qP R ¼ R qR qR qX

ð14:24Þ

with R ¼ r=Ro

X ¼ x=L

Where Ro is the lumen radius, L if the ﬁber length, u is the longitudinal convective velocity, P is pressure, X is the dimensionless axial space coordinate, R is the dimensionless radial coordinate. In writing Equation 14.24 it was assumed that ﬂow is laminar and that entrance effects can be ignored. In addition, the axial stress terms have been neglected since the aspect ratio of the hollow ﬁber (Ro/L) is typically less than 0.01. The inertial terms have been neglected also, which is valid if the radial Reynolds number (ReR ¼ ruRo/m) is much less than 1 [11]. The boundary conditions for the solution: qu ¼ 0; qR u ¼ 0;

u ¼ 0 at R ¼ 0 uo ¼ Lp DP at R ¼ 1

ð14:25aÞ ð14:25bÞ

The value DP can change in the axial direction in the hollow ﬁber (DP is the pressure drop in the membrane matrix due to the momentum transfer, the velocity through the membrane is uoe, where e is the membrane porosity). Kelsey et al. [11] have solved the equation system in all three cases, namely for closed-shell operation, partial ultraﬁltration and complete ultraﬁltration and have plotted the dimensionless axial and radial velocities as well as the ﬂow streamlines. Typical axial and radial velocity proﬁles are shown in the hollow-ﬁber membrane bioreactor at several axial positions in Figure 14.8 plotted by Kelsey et al. [11]. This ﬁgure illustrates clearly the change of the relative values of both the axial and the radial velocity [V ¼ uL/(uoRo), U ¼ u/uo where uo is the inlet centerline axial velocity]. 14.4.4.2 With Cake and Polarization Layer The principle of the calculation of the ﬂow proﬁles is the same as in the previous section. Here, the increased mass-transfer resistance should be taken into account in

14.4 Mathematical Modeling of Membrane Bioreactor

Figure 14.8 Axial (top panel and radial (bottom panel) velocity profiles in the hollow-fiber membrane bioreactor at several axial position for a 1 þ d/Ro ¼ 1.7; b/a ¼ 1.4 (where b ¼ RS/ (Ro þ d), RS is the shell radius); k Lp L2 =R3o ¼ 0:1 (k is the dimensionless permeability, Lp is the membrane permeability), f ¼ 0 (f is the retention factor). U ¼ u/uo where uo is the inlet centerline axial velocity; V ¼ uL/(uoRo).

order to calculate the pressure and the convective velocity distribution. In this case, due to the ﬁltration effect, there exists here the concentration polarization and the change of the substrate, macromolecules, microparticle concentration. This concentration change can also alter the velocity proﬁles. In this case, additionally the differential mass-balance equation should also be solved together with the momentum equations. If this is not the case then the equation system of Equations 14.23 and 14.24 can be used in this mass-transport process, as well. 14.4.5 Mass Transport in the Feed Side of the Hollow-Fiber Membrane Bioreactor

In Section 4.1, the mass transport through a membrane layer as a bioreactor was discussed. Now, in this section, the concentration distribution in the feed side of the

j325

j 14 Mathematical Modeling of Biochemical Membrane Reactors

326

membrane layer (in most cases in its lumen side) will be discussed. When the momentum equation can be regarded as independent of the mass-transport equation, these equations can be treated separately. First, we predict the convective velocities by means of the momentum equations and applying them we can predict the mass transport in the feed side of the capillary or tubular or ﬂat-plate membrane. The simpliﬁed form of the differential mass-balance equation, for example, for cylindrical coordinates, is as follows [13, 16, 57, 58]: 2 q c 1 qc qc qc þ þu ¼ u ð14:26Þ DL qr 2 r qr qr qx where r is the radial space coordinate, x is the axial space coordinate, u is the axial convective velocity, u is the radial convective velocity. When the laminar ﬂow is perfectly developed in the lumen, the value of the axial convective velocity can be given as follows: u ¼ 2umax ð1½r=Ro 2 Þ

ð14:27Þ

The axial diffusion term can often be neglected, because the convective velocity can be much higher than the axial diffusion ﬂow. The boundary conditions are as follows: c ¼ cin at x ¼ 0 for all r

ð14:28aÞ

qc ¼ 0 at r ¼ 0 for all x qr

ð14:28bÞ

qc dc ¼ um cDm uL cDL qr r¼R-o dr r¼Roþ

ð14:28cÞ

where D, Dm as well as uL, um are the diffusion coefﬁcients in the feed ﬂuid as well as the radial convective velocities in the ﬂuid and in the membrane, respectively. Ro is the lumen radius. The overall mass-transfer rates on both sides of the membrane can only be calculated when we know the convective velocity through the membrane layer. For this, Equation 14.2 should be solved. Its solution for constant parameters and for ﬁrst-order and zero-order reaction have been given by Nagy [68]. The differential equation 14.26 with the boundary conditions (14.28a) to (14.28c) can only be solved numerically. The boundary condition (14.28c) can cause strong nonlinearity because of the space coordinate and/or concentration-dependent diffusion coefﬁcient [40, 57, 58] and transverse convective velocity [11]. In the case of an enzyme membrane reactor, the radial convective velocity can often be neglected. Qin and Cabral [58] and Nagy and Hadik [57] discussed the concentration distribution in the lumen at different mass-transport parameters and at different Dm(c) functions in the case of uL ¼ 0, that is, without transverse convective velocity (not discussed here in detail).

14.5 Modeling of the MBR with Membrane Separation Unit

14.5 Modeling of the MBR with Membrane Separation Unit

In this group of MBR processes, the bioreaction takes place in a stirred-tank reactor and the puriﬁed liquid will be separated from the activated sludge or fermentation broth by a microﬁltration or ultraﬁltration module (Figures 14.1, 14.3 and 14.4). For modeling of this system, the mass transport and/or momentum transport may be separately described for both the liquid phase and the membrane module. After integration of these two models, the bioprocess can be calculated. Several research groups have investigated the performance of both the external [45] and the submerged MBR [27, 60–63] as well as moving-bed-bioﬁlm [64–66], where the support particles are suspended by aeration (Figure 14.9). These studies primarily investigated the fouling, the cake-layer formation on the ﬁltration membrane and its effect on the permeate velocity. In principle, the basic equations of the bioreactions of micro-organisms are known (see e.g., Mosers book [67]), these equations can be adapted to the moving-bed bioﬁlm membrane reactor or to the aerated bioreactor with immersed membrane or with external membrane module. 14.5.1 Moving-Bed-Biofilm Membrane Reactor

The use of immobilized biocatalyst in large-scale industrial processes is presently a widespread technique. Thus, the description of the mass transport in spherical particles as support material is well known. Ferreira et al. [32] summarizes the most important variables and equations for the mass transport, the effect of the reaction rate on the concentration distribution in the spherical particles. 14.5.2 Submerged or External MBR Process

In this case the ﬂuid phase is aerated (in the case of aerobic bioreactor) that maintains the turbulent hydrodynamic conditions on the one hand, and prevents the forming of the cake layer on the immersed membrane module, on the other hand. The reactor description is also well known [67], and is not discussed here.

Figure 14.9 Schematic representation of a moving-bed biofilm membrane reactor.

j327

j 14 Mathematical Modeling of Biochemical Membrane Reactors

328

14.5.3 Fouling in Submerged Membrane Module

The submerged membrane bioreactor is applied mostly in wastewater treatment. The activated sludge can often cause deposition of its soluble and particulate materials onto and into the membrane because of their relatively high concentration, during the wastewater treatment. Le-Clech et al. [59] discussed in detail, the fouling mechanism in their review and the different properties, namely that of membrane characteristic, biomass characteristic, operation conditions that could lead to forming a cake layer on the membrane surface. The analysis of this fouling process is not a task of this paper. The effect of the fouling on the permeate velocity can be characterized by the equations given in Section 4.3.

14.6 Conclusions and Future Prospects

In recent years, membrane bioreactors, bioreactors combined with membrane separation unit have established themselves as an alternative conﬁguration for traditional bioreactors. The important advantages offered by membrane bioreactors are the several different types of membrane modules, membrane structures, materials commercially available. Membrane bioreactors seem particularly suited to carry out complex enzymatic/microbial reactions and/or to separate, in situ, the product in order to increase the reaction efﬁciency. The membrane bioreactor is a new generation of the biochemical/chemical reactors that offer a wide variety of applications for producing new chemical compounds, for treatment of wastewater, and so on. Some topics regarding the membrane bioreactor technologies are listed here to where research efforts may be concentrated: .

. . . . .

Developing new, more effective membrane modules, and membrane material with the desired membrane structure that have narrow pore-size distribution and thus, better selectivity; Production of membrane layer with thinner skin layer; Producing charged membrane reducing signiﬁcantly the fouling during the ﬁltration of macromolecules and/or particles; Developing new membrane processes and hybrid processes in order to increase the efﬁciency of the biochemical reactions and the ﬁltration; Producing more effective, and more stabile, heat-resisting enzyme biocatalyst; Improving the microbial activities, for example, by gene manipulation.

The modeling of membrane bioreactors is in the initial stage. There are not available more or less sophisticated mathematical tools to describe the complex biochemical processes. It is not known how the mass-transport parameters, diffusion coefﬁcients, convective velocity, biological kinetic parameters might vary in function of the operating conditions, of the biolayer (enzyme/micro-organism membrane layer)

Appendix A

properties. Widespread research activities are needed to establish correlations able to produce the values of mass-transport parameters and reaction kinetic parameters.

Acknowledgement

This work was supported by the Hungarian Research Foundation under Grants OTKA 63615/2006. Appendix A

The differential mass-balance equation for state-state conditions, for cylindrical coordinate and for the mth sublayer is Dm

d2 c 1 dc dc c þ ¼0 um rmax dr 2 rm dr dr KMi þ cm

ð14:A1Þ

Rearranging Equation 14.A1 the following equation is obtained to be solved: d2 c 1 dc Pem km R2o c ¼ 0 dR2 Rm dR

ð14:A2Þ

where km ¼

vmax KMi þ cm

Pem ¼

um R o Dm

The solution of Equation 14.A2 is the same as Equation 14.6: ~

Cm ¼ Tm eðlm RÞ þ Pm eðlm RÞ with

~ l m

¼

lm

¼

Rm1 < R < Rm

ð14:A3Þ

sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 2 1 1 Pem þ Pem þ 4Ha2 Rm Rm 2 ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ s 2 1 1 Pem Pem þ 4Ha2 Rm Rm 2

as well as

sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ km R2o Ha ¼ Dm

where Ro is the radius of the cylindrical membrane, R ¼ 1 þ m(Ro þ d)/Ro, d is the membrane thickness, R ¼ r/Ro, Cm ¼ cm/cin, cin is the inlet concentration. The determination of Tm, Pm parameters is the same as in the case of Cartesian coordinates given by in Appendix B.

j329

j 14 Mathematical Modeling of Biochemical Membrane Reactors

330

Appendix B

Applying the boundary conditions, deﬁned by Equations 14.3a, 14.3b, 14.11a, 14.11b, an algebraic equation system can be obtained. The form of these equations is the same as are given here by Equation 14.B1 to Equation 14.10. The value of the ~m (with m ¼ 1,2, . . ., M) is deﬁned by parameters, am, ~am , bm, ~bm , cm, ~c m , dm, ~dm , Sm, S the boundary conditions. The expressions of these parameters are easy to get from the boundary conditions (they are not shown here). The general form of the algebraic equation system to be solved can be given as follows: S0 ¼ a1 T1 þ ~a1 P1

ð14:B1Þ

S1 ¼ b1 T1 þ ~b1 P1 þ a2 T2 þ ~a2 P2

ð14:B2Þ

~1 ¼ c1 T1 þ ~c 1 P1 þ b2 T2 þ ~b2 P2 S

ð14:B3Þ

S2 ¼ c2 T2 þ ~c 2 P2 þ a3 T3 þ ~a3 P3

ð14:B4Þ

~2 ¼ d2 T2 þ ~d2 P2 þ b3 T3 þ ~b3 P3 S ...

ð14:B5Þ

Sm ¼ cm Tm þ ~c m Pm þ am þ 1 Tm þ 1 þ ~ am þ 1 Pm þ 1

ð14:B6Þ

~m ¼ dm Tm þ ~dm Pm þ bm þ 1 Tm þ 1 þ ~bm þ 1 Pm þ 1 S .. .

ð14:B7Þ

SM1 ¼ cM1 TM1 þ ~c M1 PM1 þ aM TM þ ~ aM PM

ð14:B8Þ

~M1 ¼ dM1 TM1 þ ~dM1 PM1 þ bM TM þ ~bM PM S

ð14:B9Þ

SM ¼ cM TM þ ~c M PM

ð14:B10Þ

The value of T1 can be obtained by solving Equation 14.B11, where Chart 1 means the solution of the determinant obtained with the parameters of the of the right hand side of Equations 14.B1–14.B10 and Chart 2 is the solution of the determinant obtained from Chart 1 replacing its ﬁrst column with the left-hand side of Equations 14.B1–14.B10, according to the well-known Cramer role [69]. T1 ¼

Chart 2 Chart 1

ð14:B11Þ

Applying the well-known Cramer roles for the solution of a determinant, the value of T1 can be obtained as follows: T1 ¼ with

RM c1 EM1 aM

ð14:B12Þ

Appendix B

a1 ¼

aM ~aM cM ~c M

b1 ¼

R1 ¼ SM1 SM

bM ~bM cM ~c M ~ ~M1 SM bM Q1 ¼ S ~c M

~aM ~c M

am ¼

~aM þ 1m aM þ 1m dM þ 1m Em1 ~dM þ 1m Bm1

m ¼ 2; 3; . . . ; M

bm ¼

~bM þ 1m bM þ 1m dM þ 1m Em1 ~dM þ 1m Bm1

m ¼ 2; 3; . . . ; M1

Em ¼

cMm am dMm bm

Rm ¼ SMm

Bm ¼

~aM þ 1m

~dM þ 1m Bm1

~Mm Qm ¼ S

~bM þ 1m

~dM þ 1m Bm1

~c Mm am ~dMm bm

Rm1 Qm1

m ¼ 1; 2; . . . ; M1 am1 bm1

am1 Rm1 Qm1 bm1

m ¼ 2; 3; . . . ; M1 m ¼ 2; 3; . . . ; M1

as well as ~a1 aM1 RM ¼ S 0 RM1 QM1 ~c 1 BM1 bM1 aM ¼

~a1 a1 c1 EM1 ~c1 BM1

The solution methodology of the determinants is similar to that of the well-known Thomas algorithm used for the numerical solution of a differential equation with the ﬁnite-difference method [50]. An essential difference from the Thomas algorithm is that the ﬁrst step of the algorithm here is a so-called backward process. This means that the calculation of T1 starts from the last sublayer, that is, from the Mth sublayer of the determinant and it is continued down to the 1st sublayer. Thus, the value of T1 is obtained directly, in the ﬁst calculation step. Then, applying the known value of T1, the value of P1 can be obtained by means of the ﬁst boundary condition at X ¼ 0, namely: P1 ¼

S0 a1 T1 ~ a1

ð14:B13Þ

The values of Tm and Pm can then be obtained by means of the internal boundary conditions (m ¼ 2, . . ., M) with a forward sweep. Thus, the values of Tm þ 1 and Pm þ 1 should be determined by Equations (14.B6) and (14.B7). Thus, the values of Tm þ 1 and Pm þ 1 can be easily obtained as follows, with values of m ¼ 2, 3, . . ., M 1:

j331

j 14 Mathematical Modeling of Biochemical Membrane Reactors

332

Tm þ 1 ¼

~ m ~am þ 1 hm ~bm þ 1 h Ym þ 1

ð14:B14Þ

Pm þ 1 ¼

~ m bm þ 1 hm am þ 1 h Ym þ 1

ð14:B15Þ

with hm ¼ Sm cm Tm ~c m Pm

ð14:B16Þ

~m dm Tm ~dm Pm ~m ¼ S h

ð14:B17Þ

as well as Ym þ 1 ¼ am þ 1 ~bm þ 1 bm þ 1 ~am þ 1

ð14:B18Þ

Appendix C

The mass and momentum equations, that is, the Navier–Stokes approximation expressed in cylindrical coordinates with axisymmetry assumption, are [50, 51]: 1q qu ðruÞ þ ¼0 r qr qx qu qu qP q 1 qðruÞ q2 u r u þu ¼ þm þ 2 qr qx qx qr r qr qx qu qu qP q 1 qðruÞ q2 u u r u þu ¼ þm þ 2 2 qr qx qx qr r qr qx r

ð14:C1Þ ð14:C2Þ

ð14:C3Þ

where u is the longitudinal convective velocity, u is the radial convective velocity, r is the ﬂuid density, m is the ﬂuid viscosity, P is the pressure, x is the axial space coordinate.

References 1 Strathmann, H., Giorno, L. and Drioli, E. (2006) An Introduction to Membrane Science Technology, Institute of Membrane Technology, Italy, University of Calabria, Rende (CS). 2 Yang, W., Cicek, N. and Ilg, J. (2006) Journal of Membrane Science, 270, 201–211. 3 Giorno, L. and Drioli, E. (2000) Trends in Biotechnology, 18 (8), 339–349.

4 Marcano, J.G.S. and Tsotsis, T.T. (2002) Catalytic Membranes and Membrane Reactors, Wiley-VCH, Weinheim. 5 Brotherton, J.D. and Chau, P.C. (1990) Biotechnology and Bioengineering, 35, 375–394. 6 Sheldon, M.S. and Small, H.J. (2005) Journal of Membrane Science, 263, 30–37.

References 7 Rios, G.M., Belleville, M.-P. and PaolucciJeanjean, D. (2007) Trends in Biotechnology, 25, 242–246. 8 Charcosset, C. (2006) Biotechnology Advances, 24, 482–492. 9 Frazeres, D.M.F. and Cabral, J.M.S. (2001) Enzymatic membrane reactors, in Multiphase Bioreactor Design (eds J.M.S. Cabral, M. Mota and J. Tramper), Taylor and Francis, pp. 135–184, London. 10 Belfort, G. (1989) Biotechnology and Bioengineering, 33, 1047–1066. 11 Kelsey, L.J., Pillarella, M.R. and Zydney, A.L. (1990) Chemical Engineering Science, 45 (11), 3211–3220. 12 Schonberg, J.A. and Belfort, G. (1987) Biotechnology Progress, 3 (2), 81–89. 13 Piret, J.M. and Cooney, C.L. (1991) Biotechnology and Bioengineering, 37, 80–92. 14 Sardonini, C.A. and DiBiasio, D. (1992) Biotechnology and Bioengineering, 40, 1233–1242. 15 Godongwana, B., Sheldon, M.S. and Solomons, D.M. (2007) Journal of Membrane Science, 303, 86–99. 16 Calabro, V., Curcio, S. and Iorio, G. (2002) Journal of Membrane Science, 206, 217–241. 17 Nakajima, M. and Cardoso, J.P. (1989) Biotechnology and Bioengineering, 33, 856. 18 Giorno, L., De Bartolo, L. and Drioli, E. (2003) Membrane bioreactors for biotechnology and medical applications, in New Insight into Membrane Science and Technology: Polymeric and Bifunctional Membranes (eds D. Bhattacharyya and D.A. Butterﬁeld), Elsevier, Chapter 9, Oxford. 19 Habulin, M. and Knez, Z. (1991) Journal of Membrane Science, 61, 315–324. 20 Lozano, P., Perez-Marin, A.B., De Diego, T., Gomez, D., Paolucci-Jeajean, D., Belleville, M.P., Rios, G.M. and Iborra, J.L. (2002) Journal of Membrane Science, 201, 55–64. 21 Chung, T.P., Wu, P.C. and Jung, R.S. (2005) Journal of Membrane Science, 258, 55–63.

22 Melo, L.F. and Oliveira, R. (2001) Bioﬁlm reactors, in Multiphase Bioreactor Design (eds J.M.S. Cabral, M. Mota and J. Tramper), Taylor and Francis, pp. 271–308, London. 23 Marrot, B., Barnos-Martinez, A., Moulin, P. and Roche, N. (2004) Environmental Progress, 23, 59–68. 24 Mavrov, V. and Belires, E. (2000) Desalination, 131, 75–86. 25 Chmiel, H., Kaschek, M., Bl€ocher, C., Noronha, M. and Mavrov, V. (2003) Densalination, 152, 307–314. 26 Sarkar, B., Chakrabarti, P.P., Vijaykumar, A. and Kale, V. (2006) Desalination, 195, 141–152. 27 Rosenberger, S., Kr€ uger, U., Witzig, R., Manz, W., Szewzyk, U. and Kraume, M. (2002) Water Research, 36, 413–420. 28 Li, X.-y. and Wang, X.-m. (2006) Journal of Membrane Science, 278, 151–161. 29 Bacchin, P., Si-Hassen, D., Starov, V., Clifton, M.J. and Aimar, P. (2002) Chemical Engineering Science, 57, 77–91. 30 Song, L. (1998) Journal of Membrane Science, 144, 173–185. 31 Defrance, L. and Jaffrin, M.Y. (1999) Journal of Membrane Science, 152, 203–210. 32 Ferreira, B.S., Fernandes, P. and Cabral, J.M.S. (2001) Design and modeling of immobilized biocatalytic reactors, in Multiphase Bioreactor Design (eds J.M.S. Cabral, M. Mota and J. Tramper), Taylor and Francis, London, UK, pp. 85–180. 33 Long, S.W., Bhatia, S. and Kamaruddin, A. (2003) Journal of Membrane Science, 219, 69–88. 34 Hossain, M.M. and Do, D.D. (1989) Biotechnology and Bioengineering, 33, 963–975. 35 Salzman, G., Tadmor, R., Guzy, S., Sideman, S. and Lotan, N. (1999) Chemical Engineering Progress, 38, 289–299. 36 Waterland, L.R., Robertson, C.R. and Michaelis, A.S. (1975) Chemical Engineering Communications, 2, 37–47.

j333

j 14 Mathematical Modeling of Biochemical Membrane Reactors

334

37 Cabral, J.M.S. and Tramper, J. (1994) Bioreactor Design, in Applied Biocatalysis (eds J.M.S. Cabral, D. Best, L. Boross and J. Tramper), Harwood Academic Publishers, Switzerland, pp. 330–370. 38 Brotherton, J.D. and Chau, P.C. (1996) Biotechnology Progress, 12 (5), 575–590. 39 Lu, S.G., Imai, T., Ukita, M., Sekine, M., Higouchi, T. and Fukagawa, M. (2001) Water Research, 35 (8), 2038–2048. 40 Nagy, E. (2006) Journal of Membrane Science, 274, 159–168. 41 Nagy, E. (1999) Diffusion mass transfer in enzyme membrane reactor, in Integration of Membrane Processes into Bioconversions (eds K. Belaﬁ-Bakó, L. Gubicza and M. Mulder), Kluwer Academic/Plenum Publishers, New York, pp. 211–221. 42 Mulder, M.H.V. (1995) Polarization phenomena and membrane fouling, in Membrane Separation Technology, Principles and Applications (eds R.D. Noble and I. Stern), Elsevier, Oxford. 43 Song, L. (1999) Journal of Colloid and Interface Science, 214, 251–263. 44 Bruining, W.J. (1989) Chemical Engineering Science, 44, 1441–1447. 45 Ognier, S., Wisniewski, C. and Grasmick, A. (2004) Journal of Membrane Science, 229, 171–177. 46 Yeh, H.-M. (2002) Desalination, 145, 153–157. 47 Posh, C. and Schiewer, S. (2006) Journal of Membrane Science, 280, 284–297. 48 Bacchin, P., Hassen, D.S., Stratov, V., Clifton, M.J. and Aimar, P. (2002) Chemical Engineering Science, 57, 77–91. 49 Davis, R.H. and Scherwood, J.D. (1990) Chemical Engineering Science, 45, 3203–3209. 50 De, S. and Bhattacharya, P.K. (1997) Journal of Membrane Science, 136, 57–69. 51 Cebeci, T., Shao, J.P., Kafyeke, F. and Laurendeau, E. (2005) Computational Fluid Dynamics for Engineers,

52

53

54 55

56 57 58 59

60 61

62

63 64 65

66 67 68 69

Horizons Publishing Inc. and Springer, Heidelberg. Damak, K., Ayadi, A., Zeghamati, B. and Schmitz, P. (2004) Desalination, 161, 67–77. Song, L. and Elimelech, M. (1995) Journal of the Chemical Society-Faraday Transactions, 91 (19), 3389–3398. Kim, J.-I. and Stroeve, P. (1988) Chemical Engineering Science, 43, 247–257. Zhang, M. and Song, L. (2000) Journal of Environmental Engineering, 126, 667–674. Park, J.K. and Chang, H.N. (1986) AIChE Journal, 32, 1937–1974. Nagy, E. and Hadik, P. (2002) Desalination, 145, 147–152. Qin, Y. and Cabral, J.M.S. (1998) AIChE Journal, 41, 836–848. Le-Clech, P., Chen, V. and Fane, T.A.G. (2006) Journal of Membrane Science, 284, 17–53. Al-Malack, M.A. (2007) Desalination, 214, 112–127. Wintgens, T., Rosen, J., Melin, T., Brepols, C., Drensla, K. and Engelhardt, N. (2003) Journal of Membrane Science, 216, 55–65. Busch, J., Cruse, A. and Marquardt, W. (2007) Journal of Membrane Science, 288, 94–111. Ng, A.N.L. and Kim, A.S. (2007) Desalination, 212, 261–328. Leiknes, T. and Odeggard, H. (2007) Desalination, 202, 135–143. Artiga, P., Oyanedel, V., Garrido, J.M. and Mendez, R. (2005) Desalination, 179, 171–179. Sheldon, M.S. and Small, H.J. (2005) Journal of Membrane Science, 263, 30–37. Moser, A. (1988) Bioprocess Technology, Springer, Wien. Nagy, E. (2008) Asia-Pac. J. Chem. Eng., doi:10.1002/apj.242. Nagy, E. (2008) Chem. Eng. Res. Des., 86, 723–730.

j335

15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds Raffaele Molinari, Angela Caruso, and Leonardo Palmisano

15.1 Introduction

The principles of green chemistry are based on the necessity to develop chemical products and industrial processes that reduce or eliminate the use and the generation of toxic substances along with the risk for the human health and for the environment. In this context photocatalytic processes in membrane reactors represent a technology of great scientiﬁc interest because they allow chemical reactions and separation process to be obtained in one step, minimizing environmental and economic impacts. Heterogeneous photocatalysis is a technology that has been extensively studied for about three decades, since Fujishima and Honda discovered the photocatalytic splitting of water on TiO2 electrodes in 1972. Photocatalysis includes a large variety of reactions such us partial or total oxidations, hydrogen transfer, functionalization, rearrangements, dehydrogenation, mineralization, and so on [1]. These processes, included in a special class of oxidation techniques deﬁned as advanced oxidation processes (AOPs), are based on the irradiation of a semiconductor photocatalyst with UV light that leads to the formation of highly reactive hydroxyl radicals. In particular, water treatment in principle constitutes one of the most important ﬁelds of application of the photocatalytic processes that involve both degradation reactions for the removal of environmental pollutants and selective reactions for the synthesis of organic compounds. Coupling a membrane process to this technology it is possible to obtain the separation of the clariﬁed solution or the reaction product and also the recovery and the reuse of the catalyst. In fact, the choice of an appropriate membrane allows a selective separation of the product to be performed and to maintain the catalyst in the reaction environment in a continuous process that increases the efﬁciency of the whole system.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

336

Moreover, the use of solar light as a source of activating radiation represents an interesting future perspective on which many studies are focused.

15.2 Fundamentals on Heterogeneous Photocatalysis

Due to its high efﬁciency and the generation of harmless products, The heterogeneous photocatalytic process has caught the attention of many researchers all over the world. In recent years many studies have been realized with the purpose to understand the fundamentals of photocatalytic mechanisms and to increase the application ﬁelds [2]. 15.2.1 Mechanism

The electronic structure of a semiconductor is characterized by a ﬁlled conduction band and an empty valence band separated by a bandgap of energy (EG). When the catalyst is illuminated with photons whose energy is equal to or greater than this bandgap, the promotion of an electron from the valence band to the conduction band þ occurs with the creation of electron–hole pairs (e cb hvb ). The valence-band hole can oxidize electron-donor molecules (water or hydroxyl ions) to produce oxidizing hydroxyl radicals, whereas the conduction-band electron can reduce acceptor molecules such as O2 (to yield a superoxide ion) or a metal ion (reduced to its lower valence states) (Figure 15.1). The lowest energy level of the conduction band deﬁnes the reduction potential of the photoelectrons, while the highest one of the valence band determines the oxidizing power of the photoholes, respectively. When the reagents spread on the catalyst surface they are adsorbed on the active site and they can participate in redox reactions. The adsorbed molecule can be reduced if its reduction potential is higher than that of the photoelectrons or it can be oxidized if its potential is lower than that of the photoholes. The photonic excitation of the catalyst represents the initial step of the activation of the photocatalytic process and the hydroxyl radicals are the primary oxidant in these þ systems, although the recombination of the e cb hvb , that produces thermal energy, can occur, with a reduction of photocatalytic activity. 15.2.2 Photocatalysts: Properties and New Semiconductor Materials Used for Photocatalytic Processes

A semiconductor must exhibit some characteristics such as suitable bandgap energies, chemical and physical stability, nontoxic nature, availability and low cost, and physical characteristics that allow it to operate as photocatalyst.

15.2 Fundamentals on Heterogeneous Photocatalysis

Figure 15.1 Energy bandgap of a semiconductor.

In the literature many semiconductor materials acting as catalyst are used in photocatalytic reactions. The redox potentials of the valence and conduction bands for different semiconductors varies between þ 4.0 and 1.5 volts vs. the normal hydrogen electrode (NHE), respectively. Therefore, by careful selection of the photocatalyst a wide range of molecules can be converted via these processes [3]. The classic photocatalysts are generally oxides (TiO2, ZnO, ZrO2, CeO2, WO3, etc.) or sulﬁdes (CdS, ZnS, WS2, etc.). In Table 15.1 are reported a list of semiconductor Table 15.1 Band positions of some common semiconductors used for photocatalytic processes [2, 3, 6, 7, 95].

Semiconductor

Valence band (V vs. NHE)

Conduction band (V vs. NHE)

Bandgap (eV)

Bandgap wavelength (nm)

TiO2 anatase TiO2 rutile SnO2 ZnO ZnS Fe2O3 ZrO2 WO3 CdS CdSe GaAs GaP

þ3.1 þ3.1 þ4.1 þ3.0 þ1.4 þ2.6 þ4.2 þ3.0 þ2.1 þ1.6 þ1.0 þ1.3

0.1 þ0.1 þ0.3 0.2 2.3 0.4 0.8 þ0.2 0.4 0.1 0.4 1.0

3.2 3.0 3.8 3.2 3.7 2.2 5.0 2.8 2.5 1.7 1.4 2.3

387 380 318 387–390 335–336 560 460 443 496–497 729–730 886–887 539–540

j337

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

338

materials with their valence band and conduction-band positions, the bandgaps and the wavelength of radiation required to activate the catalyst. Some values can be different from other bibliographic sources. 15.2.2.1 Titanium Dioxide Among the various catalysts employed in the photocatalytic processes, the most popular and used is titanium dioxide, TiO2, thanks to its strong catalytic activity, high chemical stability in aqueous media and in a large range of pH (0–14), low cost due to the abundance of Ti (0.44% of the Earths crust) and long lifetime of electron–hole pairs. It occurs in nature in three forms: rutile (a tetragonal mineral usually of prismatic habit), anatase (a tetragonal mineral of dipyramidal habit that exhibits higher photocatalytic activity) and brookite (a rare orthorhombic mineral). The different photoactivity between anatase and rutile type is due to a difference in the energy structure. In both forms, the position of the valence band is deep, and the resulting positive holes show sufﬁcient oxidative power. Nevertheless, the conduction band in the anatase type is closer to the negative position than in the rutile type; therefore, the reducing power of anatase type is stronger than that of rutile type. Thus, in order to avoid photodamage, the rutile is the form employed in industrial applications as a pigment to provide whiteness and opacity to products such as plastics, papers, coatings, paints, inks, foods, or in cosmetic and skin-care products as a thickener. Instead, thanks to its high photoactivity, the anatase type is used in a wide range of applications as photocatalytic coatings on various substrates such as glass and ceramic tiles that can photodegrade various noxious or malodorous chemicals, smoke and cooking oil residues under low-intensity near-UV light. 15.2.2.2 Modified Photocatalysts Despite the wide range of application of the photocatalytic processes, their use at industrial level is still limited due to different reasons: .

. .

recombination of photogenerated electron–hole pairs: holes and excited electrons are unstable species that can quickly recombine within 10–100 ns releasing energy in the form of unproductive heat or photons [4, 5]; fast backward reactions or secondary reactions that lead to undesirable by-products; inability to use visible light: only a small fraction of solar light (less than 5% in the case of TiO2 anatase) can be utilized by the photocatalyst [6].

In order to resolve these deﬁciencies and to design high-efﬁciency and economical photocatalytic systems, the discovery and the development of new photocatalysts has become one of the most important topics in photocatalysis in the last years. Ni et al. [5], in a review on the developments in photocatalytic water splitting using TiO2 for hydrogen production, divided the techniques used to enhance the photocatalytic efﬁciency in two broad groups: photocatalyst modiﬁcation techniques,

15.2 Fundamentals on Heterogeneous Photocatalysis Table 15.2 Some recent photocatalysts and their applications.

Catalyst

Reaction

Year

References

Arginine-TiO2 Acridine yellow G (AYG) Membrane-W10 Doped-TiO2 ZnWO4 TaON AgGaS2 Fe-ZSM-5 Bi2S3/CdS Ni-doped ZnS Fe(III)-OH complexes Co3O4 POM Zn phthalocyanine complexes Pt, Au, Pd–doped TiO2 POM Hydrous alumina-doped TiO2 Bi3+–doped TiO2 Activated Carbon-ZnO ZnB12O20 La-, Cu-, Pt- doped WO3 POM POM Au/Fe2O3 YVO4

Selective reduction Total oxidation Oxidation Selective oxidation Photodegradation Total oxidation Hydrogen production Reduction Partial reduction Hydrogen production RedOx Photodegradation Functionalization Photodegradation Hydrogen production Bromination Reduction Reduction Degradation Degradation (ox) Selective oxidation Reduction Degradation (red) Degradation (ox) Degradation

2007 2007 2006 2006 2007 2005 2007 2007 2002 2000 2007 2007 2003 2007 2007 2007 2003 2007 2007 2005 2005 2004 2007 2007 2007

[4] [25] [17] [8] [24] [23] [26] [49] [15] [13] [21] [38] [56] [18] [9] [27] [14] [12] [22] [29] [11] [51] [28] [10] [30]

which include noble-metal loading, ion doping, catalyst sensitization, and chemical additives such as addition of electron donors or addition of carbonate salts to suppress backward reactions. Many studies on the use of modiﬁed catalysts have been carried out [7], some of the most recent photocatalysts and their applications are reported in Table 15.2. The main purpose of the international research is to inhibit charge recombination and to extend catalyst light absorption spectra to the visible region. It was demonstrated that it is possible, for example, loading the catalyst surface with noble metals such as Au, Pt, Pd, Ni, Ag [6, 8–10] reducing the possibility of electron–hole recombination or doping the catalyst with metal ions such as Fe3+, Co2+, Cu2+, Al3+ [11–14] that could expand its photoresponse into the visible region. In order to suppress the recombination of the photogenerated electron–hole pairs, some researchers [6, 15] have described the photocatalytic activity of composite photocatalysts consisting of two semiconductors. In these conﬁgurations, after absorption of a photon, the transfer of the electrons from the conduction band of the photoexcited component to that of the unexcited component occurs, leading to stable semiconductor particles with separated charges that do not

j339

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

340

undergo deactivation because transfer in the reverse direction is thermodynamically forbidden. Bonchio et al. [16] prepared a novel heterogeneous photo-oxidation catalysts by embedding polyoxotungstates in polymeric membranes achieving stable and recyclable photocatalytic systems with different and tunable properties depending on the nature of the polymeric materials. In particular, using PDMS and ﬂuorinated (PVDF and Hyﬂon) polymers hybrid photocatalysts, with high stability in aqueous phase under turnover regime and temperatures up to 50 C, were obtained [17]. Moreover in a certain number of recent studies are reported the use of complex catalysts, such as Zn phthalocyanine [18], Cu phthalocyanine [19], Cu porphyrin [20], Fe(III)-OH [21], activated-carbon-ZnO [22], or new synthesized photocatalysts, such us TaON [23], ZnWO4 [24], Acridine yellow G [25], AgGaS2 [26], polyoxometallates [27, 28], ZnB12O20 [29], YVO4 [30], which exhibit a signiﬁcant activity under visible light irradiation, as well as a retard in charge recombination, allowing a greater control of the whole photocatalytic process.

15.3 Photocatalytic Parameters

Some parameters that inﬂuence the photocatalytic process are: - Catalyst amount: The reaction rate is directly proportional to the catalyst dose, therefore optimal catalyst dosing is necessary for controlling the reaction. Nevertheless, above a certain value of mass catalyst (m) the rate of reaction becomes independent of m that corresponds to the maximum amount of catalyst in which all the surface active sites are occupied by the substrate. Moreover, when the catalyst is suspended in aqueous solution, aggregation of the catalyst particles occurs for high amounts, with a decrease of the number of surface active sites and the extent of transmitted light, due to increase of opacity and light scattering [2]. Therefore, the use of low amounts of catalyst is an useful condition not only from an economical point of view but also in order to guarantee a satisfactorily reaction efﬁciency. - pH of the aqueous solutions: The pH plays an important role in the efﬁciency of the photocatalytic process. In particular, the occurrence of aggregation phenomena involving the suspended TiO2 catalyst particles together with their precipitation has been observed at acidic values. The last aspect determines a reduction of the catalytic active sites and therefore a decrease of the catalyst activity. Besides, different pH values lead to the formation of different reaction products [2, 31], due to modiﬁcations in the ionization state of the catalyst surface. Depending on the substrates, an increase of the pH will have a positive or negative effect on their reaction rate because the hydrophilic/hydrophobic character of the catalyst changes with the pH. When TiO2 is used as catalyst the strongest attractive interactions occur at pH values around the point of zero charge (PZC) (values of PZC for TiO2 are

15.4 Applications of Photocatalysis

reported in the pH range 4.5–6.2). As observed by Bekkouche et al. [32] in the study of adsorption of phenol on TiO2, for example, the catalyst surface is negatively charged at alkaline value and a phenomenon of repulsion can be hypothesized that could explain the low adsorption values. Noguchi et al. [14] observed an increase of the rate of BrO3 reduction with TiO2 suspended in an aqueous solution by lowering the pH from 7 to 5. This behavior is attributable to an enhancement of the electrical interaction between the substrate and the positively charged surfaces of the TiO2 photocatalyst that causes an increase in the amount of adsorbed BrO3 on the photocatalyst surface. - Wavelength and light intensity: Only photons with l smaller than or equal to the absorption edge of the catalyst are effective in reaction activation. In addition, it must be checked that the reactants do not absorb light in order that the catalyst works in a truly heterogeneous catalytic regime. Besides, there is a direct proportionality between the conversion level and the radiation intensity that conﬁrms the participation of photoinduced electrical charges (holes and electrons) to the photocatalytic mechanism. - Presence of others species: The presence of others species in the reaction environment can enhance or decrease the rate of the photocatalytic process depending on the mechanism of reaction. As reported by Kavita et al. [2], small quantities of some ions, such as Cu2 þ or Mn2 þ , increase the rate of oxidation of organic compounds thanks to the inhibition of electron–hole recombination. On the other hand, co-dissolved ions like Cl, Na þ , PO43 affect the photodegradation rate because they possibly adsorb onto TiO2 surface competing with the q substrate and hamper the formation of OH radicals [31].

15.4 Applications of Photocatalysis

Photocatalytic processes, occurring on a semiconductor surface, can be carried out in various media: aqueous solutions, pure organic liquid phases or gas phase. The wide application ﬁeld of this technology includes a great variety of reactions such as partial or total oxidation, selective reduction, degradation of organic compounds, fuel synthesis (e.g., H2 production through water splitting), metal-corrosion prevention, disinfection, and so on [1, 7]. Some of the most recent applications of the photocatalytic processes present in literature are reported in Table 15.3. 15.4.1 Total Oxidations

The total degradation of organic pollutants represents one of the main applications of the photocatalysis. As result of a chain of oxidation reactions, which involve primarily

j341

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

342

Table 15.3 Applications of the photocatalytic processes.

Application

Substrate

Total oxidation

Dyes Pharmaceuticals Toxic organic compounds Pesticides Herbicides Hormones 2-propanol 4-methoxybenzyl alcohol Cyclohexane Hydrocarbons

Partial oxidation

Reduction

Redox

Aniline Carbonate Methane Benzene NH3 Herbicide 4-nitrophenol Metal ions BrO3 Nitrite Nitrate CO2 Dyes destruction p-chloronitrobenzene Dyes Cr(VI), BPA Cr(VI), dyes

Functionalization Halogenation Ciclization Thiolation Hydrogen production

Arenes, Cycloalkenes Amino acid Propene/H2S Water Ethanol/H2O

Product

References [10, 22, 30, 37, 38, 96] [39, 40] [18, 23–25, 29, 35]

Acetone, CO2 p-anisaldehyde Cyclohexanol Corresponding oxigenates Azobenzene Methane, methanol Methanol Phenol N2 4-aminophenol Noble metals Br Ammonia N2 Methane and H2 p-chloroaniline Cr(III), oxidation products Cr(III), oxidation products Halo-derivates, epoxide cicloderivates Propan-1-thiol

[33, 97] [34, 98, 99] [100] [8] [42] [43] [44] [101] [102] [11] [45, 46] [47] [103] [4] [48–52] [14] [53] [12] [63] [28] [54] [15] [21] [55]

[27, 56] [59] [60] [5, 26, 62] [9]

q the OH , all molecules present are mineralized to inorganic species: carbon to CO2, hydrogen to H2O, sulfurs to sulfates, halogens to halide ions and phosphorus to phosphates, respectively.

15.4 Applications of Photocatalysis

A large number of studies reports the use of photocatalytic reactions for the mineralization of pesticides [33], herbicides [34], toxic organic compounds [18, 35, 36], dyes [22, 37, 38], pharmaceuticals [39, 40]. 15.4.2 Selective Oxidations

Applications of photocatalytic oxidations as a green alternative synthetic route has been investigated by several authors although these reactions have always been considered as highly nonselective processes. Nevertheless, by selecting or modifying some photocatalytic parameters, such us the semiconductor surface or the wavelength, it is possible to control and modify the types of products and their distributions. Palmisano et al. [41] in a study on the selectivity of hydroxyl radical in the partial oxidation of different benzene derivatives have investigated how the substituent group affect the distribution of the hydroxylated compounds. The reported results show that the primary photocatalytic oxidation of compounds containing an electron donor group (phenol, phenylamine, etc.) leads to a selective substitution in ortho and para positions of aromatic molecules while in the presence of an electron-withdrawing group (nitrobenzene, benzoic acid, cyanobenzene, etc.) the attack of the OH radicals is nonselective, and a mixture of all the three possible isomers is obtained. Palmisano et al. [42], moreover, studied the selective oxidation of 4-methoxybenzyl alcohol to p-anisaldehyde in organic-free aqueous TiO2 suspensions, obtaining a considerable yield of 41.5% mol. The homemade photocatalysts were obtained under mild conditions and showed to be more selective than two common commercial samples, that is, TiO2 Degussa P25 and Merck. Nevertheless, although the reported ﬁndings are very intriguing in the light of the possibility to potentially synthesize ﬁne chemicals in green conditions, it should be highlighted that the initial alcohol concentration used in this work (circa 1.1 mM) is quite low in comparison with those used for typical organic syntheses. Colmenares et al. [8] reported the use of different metal-doped TiO2 systems for the gas-phase selective photo-oxidation of 2-propanol to acetone. They observed that doping the catalyst with Pd, Pt or Ag caused an increase in molar conversion as compared to bare TiO2, whereas the presence of Fe and Zr had a detrimental effect. Du et al. [43] performed a study on the selective photocatalytic oxidation of cyclohexane investigating the inﬂuence of the wavelength on the product yields. They observed that the major product in the absence of the catalyst at l < 275 nm was cyclohexanol, while the presence of the catalyst gave rise to a signiﬁcant formation of cyclohexanone under the same experimental conditions. When a proper light ﬁlter (Pyrex, l > 275 nm) was used, an almost complete selectivity was obtained (>95%). Selective photocatalytic reactions are also used for the conversion of hydrocarbons (cyclohexane, toluene, methylcyclohexane, ethylbenzene, and cumene) to their

j343

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

344

corresponding oxygenates [44], benzene to phenol [45, 46], ammonia to nitrogen over various TiO2 catalysts [47], methane to methanol [11], and so on. 15.4.3 Reduction Reactions

In order to photoreduce chemical compounds, the conduction band of the catalyst must be more negative than the reduction potential of the substrates. A potential and very attractive practical application of reduction by semiconductor photocatalysis technology is the removal of harmful toxic metals and the recovery of noble metals in wastewater. Metal species, such as Hg(II), Pb(II), Cd(II), Ag(I), Ni(II) and Cr(VI), are generally nondegradable and they are very toxic when present in the environment. Chen and Ray [48] in a study on the photocatalytic reduction of environmentally relevant metal ions on TiO2 suspensions, observed that the presence of dissolved oxygen, acting as electron acceptor, inhibits the reduction of metal ions with low reduction potentials, while the presence of organic reductants promotes photocatalytic reduction. Other authors reported the use of photocatalytic reduction to convert metal species as Cr(VI) to Cr(III) [49], Fe(VI) to Fe(III) [50], Pd(II) to Pd(0) [51], Hg(II) to Hg(0) [52]. The reductive potential of the photocatalysts has been exploited not only for the recovery of metallic ions but also for the degradation of other potentially toxic ions and molecules as bromate ions to Br [14], decomposition of nitrate to form nitrogen in water [12] or for the reductive destruction of dyes [28]. Moreover, the photocatalytic reduction has been used to convert some species in others of interest, such as nitrite to ammonia [53], or p-chloronitrobenzene to p-chloroaniline [54]. Further investigations reported the use of the redox potential of the photocatalytic processes for the simultaneous degradation of organic molecules and reduction of metallic ions. Liu et al. [21] investigated the simultaneous photocatalytic reduction of Cr(VI) and oxidation of bisphenol A (BPA) in an aqueous solution in the presence of Fe(III)–OH complexes as catalysts, achieving a synergy effect of the simultaneous photocatalytic oxidation and reduction of both pollutants. Papadam et al. [55], instead, coupled the reduction of Cr(VI) to the oxidative degradation of an azodye, while in another study it was reported the simultaneous photocatalytic reduction of Fe(VI) and oxidation of ammonia [50]. 15.4.4 Functionalization

By careful selection of the semiconductors it is possible to use heterogeneous photocatalysis as an alternative approach to more conventional synthetic pathways. In a study on the CH2Cl2-assisted functionalization of cycloalkenes Maldotti et al. [56] proposed the use of photoexcited decatungstate (POM) for the oxidation of

15.4 Applications of Photocatalysis

cyclohexene and cyclooctene in highly reactive chloro-intermediates able to induce mono-oxygenation and/or chlorination of alkenes in mild temperature and pressure conditions. The photocatalytic ability of POM to induce bromide-assisted functionalization process was also studied by Molinari et al. [27] in the bromination of some aromatics and alkenes. They reported the possibility to convert phenol and anisole to the corresponding monobrominated derivates and a wide range of cycloalkenes to dibromides and bromohydrins, the last ones as intermediates for the formation of epoxides. The obtained results are of great interest in ﬁne chemistry and organic synthesis, since epoxides are very versatile building blocks and halogenation of alkenes is still carried out using hazardous reagents and drastic conditions. Caronna et al. [57, 58] report the sunlight induced reactions of some heterocyclic bases (quinoline, quinaldine, quinoxaline, etc.) with amides or ethers in the presence of polycrystalline TiO2. It was found that the photoreactions occurred in heterogeneous system with higher yields than in homogenous systems under the same experimental conditions. The wide potentiality of semiconductor-mediated photocatalysis was also applied for the transformation of functional groups such as selective cyclization of amino acids in aqueous suspensions [59]. In the ﬁeld of thiochemistry, the photocatalytic synthesis of mercaptans represents an interesting chemical route. Schoumacker et al. [60] performed the synthesis of propan-1-thiol by addition of H2S on propene in contact with illuminated TiO2 or CdS catalysts, according to a reaction mechanism implying photoq generated SH radicals. 15.4.5 Hydrogen Production

Photocatalytic water splitting, a reaction in which water molecules are reduced by the electrons to form H2 and oxidized by the holes to form O2, using semiconductor materials is one of the most important reactions for solving energy and environmental problems. Hydrogen is considered as an ideal fuel for the future and its synthesis from clean and renewable energy sources represents the key component in sustainable energy systems. However, as observed by Ni et al. [5] in a review on photocatalytic water splitting for hydrogen production, presently only about 5% of the commercial hydrogen is produced primarily via water electrolysis, while other 95% is mainly derived from fossil fuels, such as natural gas, petroleum and coal. The early work of photoelectrochemical hydrogen production using TiO2 as catalyst, was reported by Fujishima and Honda [61]. Subsequently, the interest for the photocatalytic processes has grown signiﬁcantly, although the number of the reported photocatalysts used for water splitting is still limited. However, recent studies demonstrate that materials, such as Pt/SrTiO3 codoped with Cr and Sb or Ta, Pt/NaInS2, Pt/AgInZn7S9 and Cu- or Ni-doped ZnS

j345

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

346

photocatalysts, showed high activities for H2 evolution from aqueous solutions under visible light irradiation [62]. Another approach for H2 photocatalytic production is reported by Mizukoshi et al. [9] that described the syntheses of noble metal nanoparticles TiO2 composite photocatalysts by the sonochemical reduction method for hydrogen evolution from ethanol aqueous solutions at room temperature. Besides, in an interesting study Sing Tan et al. [63] reported the possibilities to use the photocatalytic reduction of carbon dioxide with water to produce hydrogen and methane. Using the photocatalytic process to reduce CO2 into hydrocarbons could contribute to the control of CO2 emission from industrial processes and both produced gases could become the key components of clean green energy systems in the future. 15.4.6 Combination of Heterogeneous Photocatalysis with Other Operations

The demand to develop efﬁcient systems for pollutant abatement and wastewater treatment has brought some researchers to develop systems in which the photocatalytic process is combined with other methods with the purpose to increase its efﬁciency and to decrease the reaction time thanks to a synergistic effect. As reported by Augugliaro et al. [64] the photocatalysis can be combined with chemical or physical operations. In the ﬁrst case, when the coupling is with ozonation [65, 66], ultrasonic irradiation, photo-Fenton reaction or electrochemical treatment, which inﬂuence the photocatalytic mechanism, an increase of the efﬁciency of the process is obtained. Coupling photocatalysis with a physical technologies, such as biological treatment [67, 68], membrane reactor [39] or physical adsorption, the combination does not affect the mechanisms but increases the efﬁciency of the whole process. The choice of the combination depends on the characteristics of the efﬂuent to be treated, in other words the best solution for all cases does not exist (especially when actual efﬂuents are under investigation), but it is mandatory to take into account all the drawbacks and the advantages before taking a decision.

15.5 Advantages and Limits of the Photocatalytic Technologies

Photocatalysis is a very promising technology that offers interesting advantages: .

The reactions occur under mild experimental conditions, that is, usually under ambient temperature and pressure without the presence of additives (only oxygen from the air) and in short times;

.

It can be applied to a wide range of substrates in aqueous, solid and gaseous phase;

.

The process can be adapted to destroy a variety of hazardous molecules and pollutants in different wastewater streams;

15.5 Advantages and Limits of the Photocatalytic Technologies .

In comparison to the simple transfer of the pollutants from one medium to another, which occurs with the conventional water-treatment methods, photocatalysis leads to the real destruction of the contaminants with the formation of innocuous products;

.

The system is applicable to solutions at low concentrations;

.

The process is able to recover noble metals and toxic metals that are converted to their less-toxic/nontoxic metallic states;

.

Combination with other physical and chemical technologies is possible;

.

It offers a good alternative to the energy-intensive conventional treatments methods with the possibility to use renewable solar energy.

Despite the great potentialities of the photocatalytic processes, their application at the industrial level is limited by different factors. Besides the problems regarding the high reactivity of the system and the low selectivity of the classical catalysts on which, as previously described, many studies have been realized in recent years, one of the main problems is the recovery of the catalyst from the reaction environment. As observed by Choi [7], with respect to the large number of studies reported on the development and synthesis of highly efﬁcient photocatalysts, few studies have been performed for the design of efﬁcient photoreactors for commercial exploitation. For the development of a continuous photocatalytic reactor, applicable at the industrial level, it is important to consider some parameters such as the catalyst conﬁguration, the speciﬁc illuminated surface area, the UV source, the mass-transfer rate and the scale-up possibilities [69]. Regarding the catalyst conﬁguration, two operating modes of the photoreactors can be identiﬁed: catalyst suspended or catalyst immobilized on a support. In the supported systems the catalyst can be coated on the walls of the reactor, supported on a solid substrate or deposited around the case of the light source. Many are the supported materials used in literature, such as glass beads, and tubes [69], silica-based materials [70], hollow beads, membranes [71], optical ﬁbers, zeolites, activated carbon, organic ﬁbers [72], and so on. In this way, there is not a requirement for separation steps, no particle aggregation occurs especially at high concentrations and the thin ﬁlms can be easily adopted to realize continuous-ﬂow systems. However, many studies [69, 73–75] showed that the suspended system is more efﬁcient. The low photocatalytic efﬁciency of the supported systems is due to several factors that inﬂuence the catalyst activity. Heterogeneous catalysis is a surface phenomenon, therefore the overall kinetic parameters are dependent on the real exposed catalyst surface area. In the supported systems only a part of the photocatalyst is accessible to light and to substrate. Besides, the immobilized catalyst suffers from the surface deactivation since the support could enhance the recombination of photogenerated electron–hole pairs and a limitation of oxygen diffusion in the deeper layers is observed. On this basis more efforts in photocatalytic engineering and reactor development are required to realize an efﬁcient photocatalytic reactor.

j347

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

348

15.6 Membrane Photoreactors 15.6.1 Introduction

The recovery of the photocatalyst from the reaction environment represents one of the main problems of the photocatalytic process that limits its industrial application. Although this process step can be obviated by the use of immobilized catalyst, the suspended system has more attractive features [76]. Therefore, the separation of the photocatalyst from the treated solution and its recycle is one of the challenges in further development of this technology. A very promising method to solve this problem is coupling the photocatalysis with membrane techniques, obtaining a very powerful process with great innovation in water treatment. In fact, membrane processes, thanks to the selective property of the membranes, have been shown to be competitive with the other separation technologies for what concerns material recovery, energy costs, reduction of the environmental impact and selective or total removal of the components [77]. The photocatalytic membrane reactors (PMRs) combine the advantages of classical photoreactors (catalyst in suspension) and those of membrane processes (separation at molecular level) with a synergy for both technologies. The membrane allows not only the recovery and reuse of the catalyst but also the selective separation of the molecules present in the reaction environment. In the conventional photoreactors the molecules and their by-products are freely transported in the ﬁnal stream giving a not efﬁcient system [73]. In a PMR, if a suitable membrane is used, it is possible to enhance the residence time of the molecules to be degraded or to obtain a selective separation of the products. Besides, the membrane photoreactors allow operation in continuous systems [39] in which the reaction of interest and the separation of the product(s) simultaneously occur, avoiding in some cases the formation of by-products. Several authors have proposed promising solutions involving the use of membranes to maintain the catalyst in suspension or to immobilize it on the membrane inside the reactor. Nevertheless, despite the potential advantages of these hybrid photoreactors, the research on the PMRs is not sufﬁciently developed yet.

15.6.2 Membrane Photoreactor Configurations

Various types of photocatalytic membrane reactors in which the catalyst was used in different modes have been built with the purpose to have an easy separation of the catalyst from the reaction environment: a photocatalyst in suspension in magnetically or mechanically agitated slurries conﬁned by means of a membrane, ﬁxed bed, catalyst deposited or entrapped on an inert support or in a membrane, and so on.

15.6 Membrane Photoreactors

15.6.2.1 Pressurized Membrane Photoreactors The majority of PMRs described in literature combines photocatalysis with a pressure-driven membrane technique, such as nanoﬁltration (NF), ultraﬁltration (UF) and microﬁltration (MF), in which the catalyst is contained in the pressurized side of the membrane. The ﬁrst studies reported in the literature on PMRs were carried out in order to identify the best techniques for conﬁning the photocatalyst, to choose suitable polymers for the membranes stable under UV-Vis illumination and to ﬁnd the inﬂuence of some photocatalytic variables on the process [77–79]. Molinari et al. [73] reported a study on various conﬁgurations of photocatalytic membrane reactors for the degradation of 4-nitrophenol using TiO2 as catalyst. Two conﬁgurations have been studied: irradiation of a recirculation tank, with the suspended catalyst conﬁned by means of the membrane, and irradiation of the cell containing the membrane with three sub-cases: catalyst deposited on the membrane, catalyst suspended, and catalyst included in a membrane during its preparation. Among the conﬁgurations described in that paper, the irradiation carried out on the recirculation batch seems very promising since it allows high irradiation efﬁciency and high membrane permeate ﬂow rate to be obtained and also it is possible to select the membrane type depending on the photocatalytic process under study. In another study, Tsuru et al. [80] reported the use of porous TiO2 membranes having pores of several nanometers for a gas-phase photocatalytic reaction of methanol as a model of volatile organic component (VOC). In this system, the titanium dioxide is immobilized in the form of a porous membrane that is capable of selective permeation and also a photocatalytic oxidation that occurs both on the surface and inside the porous TiO2 membranes. In this way, it is possible to obtain a permeate stream oxidized with OH radicals after one-pass permeation through the TiO2 membranes. Nevertheless, comparing the photodegradation efﬁciency of the membrane photoreactor made with entrapped TiO2 to that with suspended TiO2 it was observed [81] that although the amount of TiO2 used was identical, the suspended system was more efﬁcient. This is probably due to the presence of the polymer around the particles of catalyst that reduces the real amount of photoexcited TiO2. Besides, in this study it was observed that the rate of pollutant photodegradation was strongly affected by the UV irradiation mode. In particular, some experimental runs were performed by using two types of photoreactor: a cylindrical photoreactor with external lamp and an annular one with immersed lamp. Although the power of the immersed lamp was about four times lower than that of the external lamp, the ﬁrst system was three times more efﬁcient than the latter. However, since these systems need a pressure, fouling can occur with a decrease of the efﬁciency of the overall process. 15.6.2.2 Sucked (Submerged) Membrane Photoreactors With the aim to avoid or to reduce the membrane fouling, which causes the membrane ﬂux decline, another type of membrane module conﬁguration of PMR,

j349

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

350

such as the submerged membrane system has been also studied. The catalyst is suspended in an open-air reaction environment, the membrane is immersed in the batch and the permeate is sucked by means a pump. In particular, in order to control the hydrodynamic conditions near membrane surface to prevent catalyst deposition on the membrane, the research has been addressed to develop efﬁcient strategies such as gas sparging at the bottom of the membrane [82, 83]. In recent years some studies were performed using submerged membrane modules coupled to photocatalytic systems for the removal of organic pollutants such as fulvic acid [84], bisphenol-A [85], para-chlorobenzoate [86]. In a study on a submerged membrane photocatalysis reactor (SMPR) for the degradation of fulvic acid using a nanostructured TiO2 catalyst [84], the effects of the operational parameters such as TiO2 amount, pH, and airﬂow were investigated. The reported results show that the permeate ﬂux rate was improved and thus the membrane fouling was reduced with addition of nanostructured TiO2 because it has a larger particle size than P25 and therefore it can be easily separated and reused by means the membrane. Chin et al. [85] used a hybrid system combining a low-pressure submerged membrane module and a TiO2 suspension to purify water containing bisphenolA. The factors affecting the performance of a SMPR were studied and, in particular, it was observed that the aeration, allowing a mechanical agitation, reduces the fouling of the membrane and keeps the TiO2 well suspended in the solution, acting also on the size of catalyst aggregates. However, beyond an aeration of 0.5 L min1 no enhancement of photodegradation rate was observed, probably due to the presence of bubble clouds that could attenuate UV-light transmission in the photoreactor. Besides, in this study the possibility of using an intermittent permeation method was demonstrated with the aim to maintain high ﬂux (100 L h1 m2) at low aeration rate, with low membrane fouling. The ultraﬁltration or microﬁltration membrane used in the described submerged membrane photoreactors showed high ﬂuxes and a good removal efﬁciency of organic molecules, nevertheless, they are not able to reject compounds with low molecular weight. In order to solve this problem, Choi et al. [87] studied the performance of a submerged membrane bioreactor using NF cellulose acetate membrane for domestic wastewater treatment. The reported results underlined a DOC concentration in the permeates in the range 0.5–2.0 mg L1 for the ﬁrst 130 days, and subsequently increased approximately to 3.0 mg L1, indicating that the NF submerged MBR could produce very good quality permeate for a long-term operation. 15.6.2.3 Membrane Contactor Photoreactors Other types of membrane separation processes that can be useful when they are coupled with a photocatalytic system are the membrane contactors. The separation performance in these processes is determined by the distribution coefﬁcient of a component in two phases and the membrane acts only as an interface. They can be divided in gas–liquid (G–L) and liquid–liquid (L–L) membrane

15.6 Membrane Photoreactors

contactors. In the ﬁrst type, one phase is a gas or a vapor and the other phase is a liquid, whereas in the latter both phases are liquid [88]. Membrane contactors represent interesting membrane processes that allow, for example, to separate one or most reaction products from the photocatalytic environment thanks to their different distribution coefﬁcient between the two phases. In this way it is possible to solve the problem of secondary reactions that occur during a synthetic pathway and lead to undesirable by-products. Pervaporation – photocatalysis In the described systems the membrane usually permeates water and rejects the reactants, enhancing their residence time in the photoreactor. However, it is known that some intermediate products of the photocatalytic degradation of organic compounds can negatively affect the reaction rate, therefore, in some cases it is useful to eliminate these by-products in order to improve the thermodynamic and/or the kinetics of the reaction. To this purpose, in a study on the photocatalytic degradation of 4-chlorophenol, Camera-Roda and Santarelli [89] proposed an integrated system in which photocatalysis is coupled with pervaporation as process intensiﬁcation for water detoxiﬁcation. Pervaporation represents a useful separation process in the case of the removal of VOCs and in this study it is used to remove continuously and at higher rate the organic intermediates that are formed in the ﬁrst steps of the photocatalytic degradation of the weakly permeable 4-CP. Membrane distillation – photocatalysis To solve the problem of membrane fouling observed in the pressure-driven membrane photoreactor, Mozia et al. [90] studied a new type of PMR in which photocatalysis was combined with a direct contact membrane distillation (DCMD). MD can be used for the preparation of ultrapure water or for the separation and concentration of organic matter, acids and salt solutions. In the MD the feed volatile components are separated by means of a porous hydrophobic membrane thanks to a vapor-pressure difference that acts as driving force and then they are condensed in cold distillate (distilled water), whereas the nonvolatile compounds were retained on the feed side. In this study the possibility was investigated of coupling photocatalysis and membrane distillation for degradation of organic pollutants in aqueous solution using Acid Red 18 as a model dye and TiO2 Aeroxide P25 as photocatalyst. In particular, the rejection of the MD process towards the catalyst, the dye and its photodegradation products was tested and it was found that the nonvolatile model dye was completely rejected by the membrane and it remained in the feed side. In the case of TOC concentration in permeate, an initial signiﬁcant increase of concentration during the ﬁrst hour was observed, probably due to the production of some volatile by-products formed at the beginning of the process. However, these compounds can be degraded into water, CO2 and other noncarbon species during subsequent hours and, therefore, the TOC values in the permeate became lower. Dialysis – photocatalysis Azrague et al. [91] described a particular type of membrane contactor photoreactor in which a dialysis membrane (used as a contactor) was

j351

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

352

combined with a photocatalytic system for the decontamination of turbid waters. Photocatalysis is not easily applicable in the presence of solid particles that screening the radiations determine a decrease of the photocatalytic degradation. On this basis, the aim of this work was to realize a system in which the membrane process allows the solid particles to be kept in their initial compartments and also the transport of the pollutant (2,4-DHBA) by diffusion from the feed-tank compartment to the other where the photocatalytic reaction takes place, until a total mineralization, thanks to the different concentration between the two compartments. In this way, the pollutants are extracted from the turbid water and then degraded, without the need of a transmembrane pressure, avoiding the fouling of membrane, which is an expensive problem in case of pressure-driven membrane processes. 15.6.3 Parameters Influencing the Photocatalytic Membrane Reactors (PMRs) Performance

In the development of a photocatalytic membrane reactor it is important to take into account some parameters that inﬂuence the performance of the system and its applicability to the industrial level. One of the main objectives in the use of a membrane process coupled to a photocatalytic reaction is the possibility of recovering and reusing the catalyst. Moreover, when the process is used for the degradation of organic pollutants, the membrane must be able to reject the compounds and their intermediate products, while if the photocatalysis is applied to a synthesis, often the membrane have to separate the product(s) from the environment reaction. Therefore, in a PMR the choice of a suitable membrane is essential to obtain an efﬁcient system. To achieve a high membrane rejection towards the substrate it is important that the pore size of the membrane is smaller than the size of the molecules to be retained. Nevertheless, other factors inﬂuence the separation properties of a membrane, such as the shape and ﬂexibility of the substrate and its acid–base properties, as well as the concentration-polarization phenomenon and the membrane fouling. Sometimes, thanks to the Donnan effect, by modifying the pH it is possible to retain in the reaction ambient molecules that otherwise would pass the membrane. In fact, some membranes can become electrically charged at acidic and alkaline pHs, then repulsive or attractive interactions between the substrate molecules and the membrane surface may occur if the charges are of the same or of different sign, respectively. Repulsive interactions increase rejection values, whereas attractive ones decrease them. In a pressure-driven membrane process the molecules are generally rejected by the membrane and therefore their concentrations in the permeate are lower than those in the feed solution. However, an accumulation of excess particles can occur at the membrane surface with the creation of a boundary layer. This phenomenon, called concentration polarization, causes a different membrane performance. In particular, with low molecular weight solutes the observed rejection will be lower than the real retention or, sometimes, it could be negative.

15.6 Membrane Photoreactors

Moreover, in the presence of a layer deposited on the membrane surface an increase of the resistance to solvent ﬂow occurs that reduces the permeate ﬂux. This problem could be obviated by creating a turbulent ﬂow in the permeation cell, which allows catalyst and drug deposition on the membrane surface to be reduced. Another important parameter that must be take into account when a PMR is used for the water treatment is the water permeability of the membrane. The selected membrane must be able not only to selectively conﬁne the pollutants and the catalyst, but also to offer a high water permeate ﬂux in order to achieve a system for application purposes. It is worth noting that when the photocatalytic process is applied to organic synthesis, the role of the membrane becomes essential in the separation of the products from the reaction ambient. In this case it is important that the membrane is able to selectively and quickly separate the product of interest in order to avoid subsequent reactions that would cause the formation of undesirable secondary products. 15.6.4 Future Perspectives: Solar Energy

As previously described, many efforts are addressed toward the development of photocatalyst and photocatalytic systems that exploit the sun as a source of light [92, 93]. Solar energy is important for achievement of sustainable processes because it constitutes a renewable, cheap, and clean energy source. The possibility to use sunlight makes the photocatalytic membrane reactors promising in industrial and environmental ﬁelds, although very few studies have yet been performed in this area. In a study on the photodegradation of lincomycin Augugliaro et al. [40] reported the use of a hybrid system consisting of a solar photoreactor with the catalyst suspended coupled with a membrane module. The photo-oxidation experiments were performed in a batch solar photoreactor at pilot plant scale by using compound parabolic collectors (CPC), installed at the Plataforma Solar of Almerìa. By means of some preliminary tests performed without the membrane it was determined that the photo-oxidation rate of lyncomycin followed a pseudo-ﬁrst-order kinetics with respect to the substrate concentration under the used experimental conditions. The high membrane rejection values measured for lincomycin and its degradation products demonstrated that the hybrid system allowed the separation of these species and also of the photocatalyst particles, although in the experiments carried out in continuous mode, an accumulation of organic molecules in the system was observed. This ﬁnding, which was dependent on solar irradiance and initial lyncomycin concentration, can be explained by considering that the number of photons entering into the system are not sufﬁcient to mineralize the organic carbon fed into the photoreactor. Moreover, the experimental results obtained in continuous mode showed that the presence of the membrane allowed both the substrate and intermediates to be

j353

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

354

reduced down to very low concentrations, proving that the hybrid system could be very interesting from an economic point of view.

15.7 Case Study: Partial and Total Oxidation Reactions in PMRs

In this section some experimental results obtained in our laboratories on PMRs are reported. In particular two different photocatalytic membrane reactors, used in total and partial photocatalytic oxidations are described. 15.7.1 Degradation of Pharmaceutical Compounds in a PMR

The presence in the aquatic environment of pharmaceutically active compounds as an important group of toxic organic pollutants, has focused the attention of the international scientiﬁc community. The design of more stable drugs, in order to enhance their persistence in the organisms, leads to a greater resistance of these molecules to the common chemical and biological degradation treatments, with a consequent increase in the environment. The development of new systems, alternative to the traditional puriﬁcation methods (not efﬁcient and often suitable only to transfer pollutants from one phase to another) represents a great research interest. In this context, hybrid systems based on photocatalysis coupled with separation process could represent an useful solution to these problems. The aim of our experimental studies was to show the possibility to use the PMRs for the degradation of organic pollutants, in particular drugs, in water, considering different reacting system conﬁgurations of membrane photoreactors and investigating the effects of some parameters on the efﬁciency of the process. The obtained results have shown that the conﬁguration where the recirculation tank was irradiated and the catalyst was used in suspension appeared to be the most interesting for industrial applications [73]. Moreover, it was observed that the degradation rate was higher when an immersed lamp was used compared to a system with an external lamp [81]. Therefore, actually the studies in progress are realized in the system described elsewhere [39] consisting of a Pyrex annular photoreactor with a 125-W medium-pressure Hg lamp axially positioned inside the reactor. The separation module containing the ﬂat-sheet membrane was connected to the photoreactor in a recirculation loop. In a ﬁrst set of runs the inﬂuence of pH on the rejection of several pharmaceuticals was studied by using different membranes. Among the tested membranes, the NTR7410 membrane, a NF commercial membrane made of sulfonated polyethersulfone, resulted in a good compromise for the drugs studied both at acidic and alkaline pHs, with an average ﬂux of ca. 45 L h1 m2 but with low rejection values that decreased down to zero during the photodegradation (Figure 15.2).

15.7 Case Study: Partial and Total Oxidation Reactions in PMRs

Figure 15.2 Drug concentrations versus time for runs carried out by using the hybrid system with the NTR 7410 membrane at initial pH of 3 (a) and 11 (b) (CTiO2 ¼ 1 g/l; CO2 ¼ 22 ppm; immersed lamp 125 W). Furosemide: (*) retentate; (.) permeate. Ranitidine: (&) retentate; (&) permeate [39].

j355

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

356

This ﬁnding depends on different factors and, as previously described, was partially solved by enhancing the turbulent ﬂow on the membrane surface. In this way, the deposition of the substrate and the catalyst was limited, avoiding the concentration polarization phenomenon that also affects the water ﬂux across the membrane. This aspect is currently under study and some preliminary results obtained in a study on the Gemﬁbrozil degradation in the described PMR are reported in Table 15.4. In addition, the maximum beneﬁt when a photocatalytic membrane reactor is used for water puriﬁcation consists in retaining not only the pollutants but also their intermediates. By means of TOC analyzes carried out during photodegradation experiments in closed and continuous membrane photoreactors it was observed that some oxidation products pass through the membrane. Therefore, further investigation is in progress to analyze this aspect. Particular attention is addressed to the permeate ﬂux and to this purpose some preliminary experiments were realized on a different conﬁguration of membrane photoreactor with a submerged membrane module located separately from the photoreactor. Bubbled oxygen on the membrane surface has the roles to reduce the catalyst deposition, to increase the ﬂux through the membrane and to facilitate the photocatalytic reaction. The submerged membrane photoreactor was more advantageous in terms of permeate ﬂux, with values almost twice those measured with the ﬂat-sheet membranes. Nevertheless, the UF membranes used in the submerged system were not able to reject the drug and its degradation products [94]. Consequently further studies are required to look for different types of membranes, such as for instance higher rejection NF-type or low rejection reverseosmosis-type membranes, by taking into account the relatively low molecular weight of the drugs studied.

Table 15.4 Rejection values vs. TiO2 concentration measured with pump flow rate of 10 L h1 and 46 L h1 [94].

[TiO2]

Q ¼ 10 L h1 R%

Q ¼ 46 L h1 R%

0 0.01 0.02 0.04 0.06 0.10 0.5 1

75.9 56.3 48.2 38.9 31.9 20.85 / 60.0

88.25 / 86.7 / 80.5 76.7 69.8 25.9

15.7 Case Study: Partial and Total Oxidation Reactions in PMRs

15.7.2 Photocatalytic Production of Phenol from Benzene in a PMR

The main aim of this study is to prove the possibility to use a PMR for the synthesis and the separation of substances of industrial interest. In particular, our research is addressed to the simultaneous one-step production and separation of phenol by selective oxidation of benzene in a membrane photoreactor using TiO2 as catalyst and a membrane contactor for the separation process. Phenol is an important chemical intermediate of industrial interest, used as substrate for the production of antioxidants, polymers and agrochemicals. Actually, more than of 90% of the world production is realized by the three-step cumene process that leads to the formation of acetone as by-product. The one-step hydroxylation of benzene represents an attractive alternative pathway for the direct synthesis of phenol and many studies are performed using different processes among which the photocatalytic reaction [45, 46]. One of the main problem is the low selectivity of the process due to the higher reactivity of phenol towards the oxidation than benzene with the formation of oxidation by-products. In order to avoid these secondary products and to obtain the separation of the phenol from the oxidant reaction environment the use of a membrane system coupled with the photocatalytic process seems a useful solution. On this basis, our research is based on the development of an experimental system that allows both high yields and good selectivity of the process, limiting the formation of undesirable by-products, and an efﬁcient separation by the identiﬁcation of a membrane with high phenol permeability and complete rejection to the catalyst. By means of preliminary batch tests some important parameters that inﬂuence the photocatalytic oxidation of benzene to phenol were investigated. In particular, the obtained results showed an increase of phenol production depending on the pH of the aqueous TiO2 suspensions, the catalyst concentration and the radiation intensity. The membrane photoreactor under investigation consists of an external lamp placed on a batch reactor containing the aqueous solution with the catalyst in suspension; by means of a peristaltic pump the solution is withdrawn from the

Figure 15.3 Phenol concentrations in aqueous and organic phase during an experimental run in the PMR (Molinari et al., unpublished data).

j357

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

358

photocatalytic reactor to a membrane contactor module in which a benzene solution is present as a strip phase. The data obtained in a ﬁrst set of experiments have shown a good separation of phenol from the reaction environment (Figure 15.3), although also other oxidation products passed through the membrane. Therefore, further studies are in progress in order to obtain a better separation efﬁciency and a faster synthetic process.

15.8 Conclusions

Photocatalytic membrane reactors represent a very promising technology of great research and industrial interest. The combination of heterogeneous photocatalysis with membrane processes provides many advantages in terms of output and costs thanks to their synergy. It is well known that heterogeneous photocatalysis can be successfully used to photodegrade or to transform a wide range of molecules in liquid–solid and in gas–solid systems. Nevertheless, the knowledge of fundamentals of photocatalysis is essential to understand the mechanistic aspects and to ﬁnd the parameters that inﬂuence the process under investigation. Moreover, the development of new photocatalysts and their application in the various research ﬁelds is a mandatory task. Some drawbacks deriving from the use of a single technology can be minimized by coupling them. Membrane processes, indeed, thanks to their selective properties, allow not only to recovery and to reuse the photocatalyst but also to enhance the residence time of the substrates to be degraded or to obtain a selective separation of the products. Various conﬁgurations of membrane photoreactors described can be chosen to inﬂuence the performance of the photocatalytic systems and possible solutions can be found to solve some problems such as the control of the catalyst activity and the fouling, the selectivity and the rejection of the membrane. A sustainable process can be obtained when the PMRs is used exploiting Sun as a cheap and clean source of light. Work carried out in our laboratories using PMRs showed the possibility to apply them in processes for total or partial oxidation of organic compounds in water. PMRs can be considered an useful green system for water puriﬁcation as well as for organic synthesis, although additional studies are still needed before taking advantage of their potentiality at the industrial level.

References 1 Herrmann, J.-M. (2005) Topics in Catalysis, 34, 49. 2 Kavita, K., Rubina, C. and Rameshwar, L.S. (2004) Industrial &

Engineering Chemistry Research, 43, 7683. 3 Robertson, P.K.J. (1996) The Journal of Cleaner Production, 4, 203.

References 4 Ahn, W.-Y., Sheeley, S.A., Rajh, T. and Cropek, D.M. (2007) Applied Catalysis BEnvironmental, 74, 103. 5 Ni, M., Leung, M.K.H., Leung, D.Y.C. and Sumathy, K. (2007) Renewable and Sustainable Energy Reviews, 11, 401. 6 Robert, D. (2007) Catalysis Today, 122, 20. 7 Choi, W. (2006) Catalysis Surveys from Asia, 10, 16. 8 Colmenares, J.C., Aramendıa, M.A., Marinas, A., Marinas, J.M. and Urbano, F.J. (2006) Applied Catalysis A-General, 306, 120. 9 Mizukoshi, Y., Makise, Y., Shuto, T., Hu, J., Tominaga, A., Shironita, S. and Tanabe, S. (2007) Ultrasonics Sonochemistry, 14, 387. 10 Wang, C.-T. (2007) Journal of NonCrystalline Solids, 353, 1126. 11 Taylor, C.E. (2005) Topics in Catalysis, 32, 179. 12 Rengaraj, S. and Li, X.Z. (2007) Chemosphere, 66, 930. 13 Kudo, A. and Sekizawa, M. (2000) Chemical Communications, 1371. 14 Noguchi, H., Nakajima, A., Watanabe, T. and Hashimoto, K. (2003) Environmental Science & Technology, 37, 153. 15 Kobasa, I.M. and Tarasenko, G.P. (2002) Theoretical Chemistry Accounts, 38 (4), 255. 16 Bonchio, M., Carraro, M., Scorrano, G., Fontananova, E. and Drioli, E. (2003) Advanced Synthesis and Catalysis, 345, 1119. 17 Bonchio, M., Carraro, M., Gardan, M., Scorrano, G., Drioli, E. and Fontananova, E. (2006) Topics in Catalysis, 40 (1–4), 133. 18 Marais, E., Klein, R., Antunes, E. and Nyokong, T. (2007) Journal of Molecular Catalysis A-Chemical, 261, 36. 19 Mele, G., Ciccarella, G., Vasapollo, G., Garcıa-López, E., Palmisano, L. and Schiavello, M. (2002) Applied Catalysis B-Environmental, 38, 309. 20 Mele, G., Del Sole, R., Vasapollo, G., Garcıa-López, E., Palmisano, L., Jun, L., Słota, R. and Dyrda, G. (2007) Research on Chemical Intermediates, 33, 433.

21 Liu, Y., Deng, L., Chen, Y., Wu, F. and Deng, N. (2007) Journal of Hazardous Materials. B, 139, 399. 22 Sobana, N. and Swaminathan, M. (2007) Solar Energy Materials and Solar Cells, 91, 727. 23 Ito, S., Thampi, K.R., Comte, P., Liska, P. and Gr€atzel, M. (2005) Chemical Communications, 268. 24 Huang, G. and Zhu, Y. (2007) Materials Science and Engineering B, 139, 201. 25 Amat, A.M., Arques, A., Galindo, F., Miranda, M.A., Santos-Juanes, L., Vercher, R.F. and Vicente, R. (2007) Applied Catalysis B-Environmental, 73, 220. 26 Jang, J.S., Choi, S.H., Shin, N., Yu, C. and Lee, J.S. (2007) Journal of Solid State Chemistry, 180, 1110. 27 Molinari, A., Varani, G., Polo, E., Vaccari, S. and Maldotti, A. (2007) Journal of Molecular Catalysis A-Chemical, 262, 156. 28 Troupis, A., Gkika, E., Triantis, T., Hiskia, A. and Papacostantinou, E. (2007) Journal of Photochemistry and Photobiology AChemistry, 188, 272. 29 Tang, J. and Ye, J. (2005) Chemical Physics Letters, 410, 104. 30 Xu, H., Wang, H. and Yan, H. (2007) Journal of Hazardous Materials, 144, 82. 31 Robert, D. and Malato, S. (2002) The Science of the Total Environment, 291, 85. 32 Bekkouche, S., Bouhelassa, M., Hadj Salah, N. and Meghlaoui, F.Z. (2004) Desalination, 166, 355. 33 Lhomme, L., Brosillon, S. and Wolbert, D. (2007) Journal of Photochemistry and Photobiology A-Chemistry, 188, 34. 34 Singh, H.K., Saquib, M., Haque, M.M. and Muneer, M. (2007) Journal of Hazardous Materials, 142, 425. 35 Waldner, G., Br€ uger, A., Gaikwad, N.S. and Neumann-Spallart, M. (2007) Chemosphere, 67, 779. 36 Wu, C., Chang-Chien, G. and Lee, W. (2005) Journal of Hazardous Materials. B, 120, 257. 37 Chen, C.-C. (2007) Journal of Molecular Catalysis A-Chemical, 264, 82.

j359

j 15 Photocatalytic Membrane Reactors in the Conversion or Degradation of Organic Compounds

360

38 Lou, X., Han, J., Chu, W., Wang, X. and Cheng, Q. (2007) Materials Science and Engineering. B, 137, 268. 39 Molinari, R., Pirillo, F., Loddo, V. and Palmisano, L. (2006) Catalysis Today, 118, 205. 40 Augugliaro, V., Garcıa-López, E., Loddo, V., Malato-Rodrıguez, S., Maldonado, I., Marcì, G., Molinari, R. and Palmisano, L. (2005) Solar Energy, 79, 402. 41 Palmisano, G., Addamo, M., Augugliaro, V., Caronna, T., Di Paola, A., Garcıa López, E., Loddo, V., Marcì, G., Palmisano, L. and Schiavello, M. (2007a) Catalysis Today, 122, 118. 42 Palmisano, G., Yurdakal, S., Augugliaro, V., Loddo, V. and Palmisano, L. (2007b) Advanced Synthesis and Catalysis, 349, 964. 43 Du, P., Moulijin, J.A. and Mul, G. (2006) Journal of Catalysis, 238, 342. 44 Gonzales, M.A., Howell, S.G. and Sikdar, S.K. (1999) Journal of Catalysis, 183, 159. 45 Park, H. and Choi, W. (2005) Catalysis Today, 101, 291. 46 Shimizu, K., Akahane, H., Kodama, T. and Kitayama, Y. (2004) Applied Catalysis A-General, 269, 75. 47 Yamazoe, S., Okumura, T. and Tanaka, T. (2007) Catalysis Today, 120, 220. 48 Chen, D. and Ray, A.K. (2001) Chemical Engineering Science, 56, 1561. 49 Kanthasamy, R. and Larsen, S.C. (2007) Microporous and Mesoporous Materials, 100, 340. 50 Sharma, V.K. and Chenay, B.V.N. (2005) Journal of Applied Electrochemistry, 35, 775. 51 Troupis, A., Hiskia, A. and Papaconstantinou, E. (2004) Applied Catalysis B-Environmental, 52, 41. 52 Wang, X., Pehkonen, S.O. and Ray, A.K. (2004) Electrochimica Acta, 49, 1435. 53 Ranjit, K.T., Krishnamoorthy, R., Varadarajan, T.K. and Viswanathan, B. (1995) Journal of Photochemistry and Photobiology A-Chemistry, 86, 185. 54 Zhang, T., You, L. and Zhang, Y. (2006) Dyes and Pigments, 68, 95. 55 Papadam, T., Xekoukoulotakis, N.P., Poulios, I. and Mantzavinos, D. (2007)

56

57

58

59 60

61 62 63

64

65

66

67

68

69

Journal of Photochemistry and Photobiology A-Chemistry, 186, 308. Maldotti, A., Amadelli, R., Vitali, I., Borgatti, L. and Molinari, A. (2003) Journal of Molecular Catalysis A-Chemical, 204, 703. Caronna, T., Gambarotti, C., Palmisano, L., Punta, C. and Recupero, F. (2003) Chemical Communications, 2350. Caronna, T., Gambarotti, C., Palmisano, L., Punta, C. and Recupero, F. (2005) Journal of Photochemistry and Photobiology A-Chemistry, 171, 241. Ohtani, B., Pal, B. and Shigeru, I. (2003) Catalysis Surveys from Asia, 7, 165. Schoumacker, K., Geantet, C., Lacroix, M., Puzenat, E., Guillard, C. and Hermann, J.-M. (2002) Journal of Photochemistry and Photobiology A-Chemistry, 152, 147. Fujishima, A. and Honda, K. (1972) Nature, 238, 37. Kudo, A. (2003) Catalysis Surveys from Asia, 7 (1), 31. Sing Tan, S., Zou, L. and Hu, E. (2007) Science and Technology of Advanced Materials, 8, 89. Augugliaro, V., Litter, M., Palmisano, L. and Soria, J. (2006) Journal of Photochemistry and Photobiology C: Photochemistry Reviews, 7, 127. Addamo, M., Augugliaro, V., GarcıaLópez, E., Loddo, V., Marcì, G. and Palmisano, L. (2005) Catalysis Today, 107–108, 612. Hernandez-Alonso, M.D., Coronado, J.M., Maira, A.J., Soria, J., Loddo, V. and Augugliaro, V. (2002) Applied Catalysis BEnvironmental, 39, 257. Parra, S., Sarria, V., Malato, S., Peringer, P. and Pulgarin, C. (2000) Applied Catalysis B-Environmental, 27, 153. Pulgarin, C., Invernizzi, M., Parra, S., Sarria, V., Polania, R. and Peringer, P. (1999) Catalysis Today, 54, 341. Dijkstra, M.F.J., Buwalda, H., de Jong, A.F., Michorius, A., Wilkelman, J.G.M. and Beenackers, A.A.C.M. (2001) Chemical Engineering Science, 56, 547.

References 70 Coronado, J.M., Soria, J., Conesa, J.C., Bellod, R., Adan, C., Yamaoka, H., Loddo, V. and Augugliaro, V. (2005) Topics in Catalysis, 35 (3–4), 279. 71 Rincón, A.G. and Pulgarin, C. (2003) Applied Catalysis B-Environmental, 44, 263. 72 Rao, K.V.S., Subrahmanyam, M. and Boule, P. (2004) Applied Catalysis BEnvironmental, 49, 239. 73 Molinari, R., Palmisano, L., Drioli, E. and Schiavello, M. (2002) Journal of Membrane Science, 206, 399. 74 Mascolo, G., Comparelli, R., Curri, M.L., Lovecchio, G., Lopez, A. and Agostiano, A. (2007) Journal of Hazardous Materials, 142, 130. 75 Sakkas, V.A., Arabatzis, I.M., Konstantinou, I.K., Dimou, A.D., Albanis, T.A. and Falaras, P. (2004) Applied Catalysis B-Environmental, 49, 195. 76 Sopajaree, K., Qasim, S.A., Basak, S. and Rajeshwar, K. (1999) Journal of Applied Electrochemistry, 29, 533. 77 Molinari, R., Grande, C., Drioli, E., Palmisano, L. and Schiavello, M. (2001) Catalysis Today, 67, 273. 78 Sopajaree, K., Qasim, S.A., Basak, S. and Rajeshwar, K. (1999) Journal of Applied Electrochemistry, 29, 1111. 79 Molinari, R., Mungari, M., Drioli, E., Di Paola, A., Loddo, V., Palmisano, L. and Schiavello, M. (2000) Catalysis Today, 55, 71. 80 Tsuru, T., Kan-no, T., Yoshioka, T. and Asaeda, M. (2003) Catalysis Today, 82, 41. 81 Molinari, R., Pirillo, F., Falco, M., Loddo, V. and Palmisano, L. (2004) Chemical Engineering and Processing, 43, 1103. 82 Ghosh, R. (2006) Journal of Membrane Science, 274, 73. 83 Chan, C.C.V., Berube, P.R. and Hall, E.R. (2007) Journal of Membrane Science, 297, 104. 84 Fu, J., Ji, M., Wang, Z., Jin, L. and An, D. (2006) Journal of Hazardous Materials. B, 131, 238. 85 Chin, S.S., Lim, T.M., Chiang, K. and Fane, A.G. (2007) Chemical Engineering Journal, 130, 53.

86 Huang, X., Meng, Y., Liang, P. and Qian, Y. (2007) Separation and Puriﬁcation Technology, 55, 165. 87 Choi, J.H., Fukushi, K. and Yamamoto, K. (2007) Separation and Puriﬁcation Technology, 52, 470. 88 Mulder, M. (1991) Basic Principles of Membrane Technology, 2nd edn, Kluwer Academic Publishers, Dordrecht. 89 Camera-Roda, G. and Santarelli, F. (2007) Journal of Solar Energy EngineeringTransactions of the ASME, 129, 68. 90 Mozia, S., Tomaszewska, M. and Morawski, A.W. (2005) Applied Catalysis B-Environmental, 59, 131. 91 Azrague, K., Aimar, P., Benoit-Marquie, F. and Maurette, M.T. (2007) Applied Catalysis B-Environmental, 72, 197. 92 Fernandez, P., Blanco, J., Sichel, C. and Malato, S. (2005) Catalysis Today, 101, 345. 93 Sarria, V., Peringer, P., Caceres, J., Blanco, J., Malato, S. and Pulgarin, C. (2004) Energy, 29, 853. 94 Molinari, R., Caruso, A., Argurio, P. and Poerio, T. (2008) Journal of Membrane Science, 319, 54. 95 Liu, X., Li, Y. and Wang, X. (2006) Materials Letters, 60, 1943. 96 Kansal, S.K., Singh, M. and Sud, D. (2007) Journal of Hazardous Materials, 141, 581. 97 Abu Tariq, M., Faisal, M., Muneer, M. and Bahnemann, D. (2007) Journal of Molecular Catalysis A-Chemical, 265, 231. 98 Sleiman, M., Conchon, P., Ferronato, C. and Chovelon, J.M. (2007) Applied Catalysis B-Environmental, 71, 279. 99 Topalov, A., Molnar-Gabor, D., Kosanic, M. and Abramovic, B. (2000) Water Research, 34 (5), 1473. 100 Zhang, Y., Zhou, J.L. and Ning, B. (2007) Water Research, 41, 19. 101 Karunakaran, C.and Senthilvelan,S. (2006) Electrochemistry Communications, 8, 95. 102 Ku, Y., Lee, W.H. and Wang, W.Y. (2004) Journal of Molecular Catalysis A-Chemical, 212, 191. 103 Zertal, A., Molnar-Gabor, D., Malouki, M.A., Sehili, T. and Boule, P. (2004) Applied Catalysis B-Environmental, 49, 83.

j361

j363

16 Wastewater Treatment by Membrane Bioreactors TorOve Leiknes

16.1 Introduction

Water is fundamental for life, is by far the most important food item and a commodity that modern societies rely on in many aspects including potable water, agricultural water, industrial water, and recreational water. Water is essential and preservation of its safety in quantity and in quality is critical to the sustainable development of any society. Historical documents show that water has always been an important issue in all civilizations, however, sanitary engineering or environmental engineering as we know it today has a very limited history. The terms sanitary engineering, or public health engineering came about during the mid-nineteenth century when the close coupling between water quality, sanitation, and public health were discovered. Epidemic outbreaks of cholera, typhoid, and dysentery in European cities during that time initiated the steps, developments and implementation of what we understand today as environmental engineering. During this period investigations were done to try and understand the general ill health in urban areas. Although reports suggested that the public-health problems were mainly due to poor sanitation and unclean water supplies, it was not until the cholera epidemic in London in 1854 and the work by John Snow showing a direct connection between contaminated water and spreading of the disease that authorities were convinced [7]. Recommendations from these investigations were to secure proper urban sanitation, provide a clean water supply and a proper drainage system to remove human wastes. Following this revelation, the objectives of all developments within the ﬁeld of environmental engineering were primarily to provide drinking water that looked and tasted good, to prevent the waterborne diseases, and build a system for wastewater collection. Over time the objective also included protecting the natural environment from negative impacts caused by wastewater contamination. With respect to wastewater treatment, probably the most signiﬁcant development was the implementation of the activated sludge process in 1913 by Arden and Lockett [1, 3, 54]. This process is the basis of biological treatment of wastewater by intensive aeration and mixing of the suspended solids formed and has found a wide application worldwide. As such, the treatment

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 16 Wastewater Treatment by Membrane Bioreactors

364

scheme now known as the conventional activated sludge (CAS) process has been by far the most common approach and solution for municipal wastewater treatment. Today the world is running out of clean, safe, fresh water. By 2025 one third of humanity – almost three billion people – will face severe water scarcity. This has been described as the single greatest threat to health, the environment and global food security. At the Johannesburg Summit on sustainable development (SD) 2002, two of the Millennium goals were deﬁned as; by 2015 reduce by half the proportion of people without access to basic sanitation (2 billion people), and reduce by half the proportion of people without sustainable access to safe drinking water (1.5 billion people). Continuousextractionof waterhas resulted indepletion ofavailable watersourcesin andaround the industrial areas in many regions of the world. In addition, wastewater discharges into natural watercourses has caused surface and groundwater pollution, often leaving water unsafe for potable use and impairing industrial use without major and costly treatment. The current low-cost end-of-pipe treatment approach will become increasingly expensive as efﬂuent discharge standards become more stringent. Meanwhile, technological advancements nowmake itpossible to treat wastewater for a variety of end uses as direct discharge to sensitive areas, industrial or even potable-water reuse. The development of advanced wastewater-treatment strategies is necessary to implement sustainable water management in general. Efﬁcient but cost-effective wastewater-treatment processes are needed for two purposes; both for producing high-quality water from contaminated resources and for transforming wastewater into water able to be reused for various applications – for potable water, in agriculture and industry. Due to their unique characteristics, and mainly the possibility to adjust the retention efﬁciency to the level which is needed (from colloids and microorganisms to small molecules and ions), membranes will play a more and more important role in the near future. To both the problems of water management and water quality, advanced membrane technologies offer practical, cost-effective and energy-saving solutions whether for large-, medium-, or small-scale applications. Membrane-bioreactor processes represent such advanced technologies for the treatment of both municipal and industrial wastewater. Furthermore, membrane processes are suited for onsite small-scale reuse. Presently, water-piping costs represents a major part of the cost of the water and wastewater management and there is a strong tendency in this area to discuss decentralized solutions. There is also a trend towards discussing a paradigm shift in environmental engineering looking into new and alternative technologies and strategies to meet future demands. Membrane technology will have an important role in these developments.

16.2 Membranes in Wastewater Treatment 16.2.1 Background

The history of membranes applied to treatment of wastewater is relatively young, dating back to the late 1960s. Over the last 40–50 years there has been a clear evolution

16.2 Membranes in Wastewater Treatment

of what kind of membranes have been applied and what the objective of the treatment scheme has been. Membrane technology has been applied to various types of wastewater. The largest number of installations is probably for industrial wastewater applications, however, municipal wastewater is largest in volume treated. The emphasis of wastewater treatment by membranes in this chapter will be for municipal wastewater treatment. 16.2.2 Membranes Applied to Wastewater Treatment

The evolution and possible applications of membranes in wastewater treatment is illustrated with the examples shown in Figure 16.1. Production of fresh water from brackish water or saline water sources by membrane desalination using reverse osmosis (RO) was ﬁrst commercialized in the late 1950s, and in fact the starting point of membranes applied to wastewater [12, 55]. Due to the superior separation properties of the RO system and the possibility of making high-grade water, RO systems were applied as a ﬁnal step in tertiary wastewater treatment [4]. In the ﬁrst applications, they were preceded by several unit processes to ensure stable operation and minimum fouling. With the subsequent developments of MF/UF membranes, a more efﬁcient pretreatment for the RO units was possible and the conventional unit processes needed to achieve tertiary treatment standards could be replaced by the combined membrane processes. Today, ﬁnal treatment by RO as the last stage is considered state-of-the-art technology for reuse and recycling of wastewater. MF/UF as a replacement for conventional sedimentation in activated-sludge processes was ﬁrst reported in the late 1960s. The membrane sewage treatment (MST) process applied was based on an activated-sludge reactor coupled with a continuous withdrawal of water through rotating-drum screen followed by a UF unit. During this period other bench-scale studies were being conducted where MF/UF membranes in a side-stream conﬁguration were used for ﬁltration of activated sludge [10, 40, 72]. These systems were not very efﬁcient and energy demanding, however, they succeeded in demonstrating and establishing the fact that membrane technology could be coupled with the activated-sludge process for wastewater treatment. The technology ﬁrst entered the Japanese market where small systems were applied to both domestic and industrial applications. The Japanese Government subsequently joined several large companies and established programs with the aim to develop a technology with small footprint, high-quality efﬂuent that was suitable for wastewater recycling [72]. Around this time other developments around the world were taking place and in the late 1980s to early 1990s the concept of the membrane bioreactor (MBR) was established as an alternative process to the conventional activated-sludge process [88]. As illustrated in Figure 16.1(d), wastewater treatment by membrane technology is an established alternative, particularly in sensitive areas, water-scarce regions, and in cases where wastewater reuse and recycling is required. A wide variety of products are now available with an increasing number of reference sites. Conﬁdence in the technology is growing, and the implementation and growth of MBR technology at a signiﬁcant pace is expected.

j365

j 16 Wastewater Treatment by Membrane Bioreactors

366

Figure 16.1 Illustration of evolution and use of membranes in wastewater treatment.

Membrane bioreactors (MBR) are commonly understood as the combination of membrane ﬁltration and biological treatment using activated sludge (AS) where the membrane primarily serves to replace the clariﬁer in the wastewater-treatment system [18, 29, 79, 85]. A major advantage of the MBR system is that it can operate at a much higher solids concentration in the bioreactor than that of a conventional CAS – mixed liquor suspended solids (MLSS) concentrations typically in the range 8–12 kg/l compared to CAS that can only work at about 2–3 kg/l, because of the limitations on settling. This higher sludge concentration permits effective removal, not only of dissolved organic material but also of residual particulate solids. A

16.2 Membranes in Wastewater Treatment

comparison and assessment of MBR technology vs. the conventional activated-sludge process generally highlights the following issues: Improved Water Quality . .

Meets stringent efﬂuent requirements; Filters out nearly all solids. Allows Wastewater Reuse

. . .

As part of a treatment scheme, provides water for potable reuse; Reduces wastewater discharge fees and freshwater costs; Provides water for nonpotable applications where fresh water is in short supply. Lowers Capital Costs

. .

Clariﬁer is not needed; Biological step can be scaled down since bacteria concentration is higher. Reduces Plant-Space Requirements

. .

Footprint is up to 50% smaller than conventional plant; Allows for expanded capacity within existing buildings. Fewer Operational Problems

.

Bulking and ﬂoating sludge problems are avoided.

Although there are clearly some beneﬁts of the MBR process compared to CAS the technology is not void of any disadvantages. In the early years of development the process was considered to be expensive due to high membrane costs, uncertainties of membrane lifetime and anticipated membrane replacement costs. As MBR plants have been in operation for a while and experience has been gained, membrane lifetime appears to be longer than initially thought and replacement costs stipulated in the early 1990s to be 80–90% of O&M costs is now estimated to be around 10–15%. This reduction is a combination of gained conﬁdence in the technology, better and cheaper production of membrane modules and product development in general. The main disadvantage of MBR systems is membrane fouling, an inherent phenomenon in all membrane processes. Strategies and techniques to alleviate fouling coupled with the frequency of membrane cleaning is one of the main constraints of the process. The high energy demands for aeration, both for the biological process and membrane operation, is currently recognized as another major challenge and drawback of the technology. MBR technology is probably the membrane process that has had most success and has the best prospects for the future in wastewater treatment. Trends and developments also indicate that this technology is becoming accepted and is rapidly becoming the best available technology (BAT) for many wastewater-treatment applications. The cost of an MBR plant for secondary treatment is still higher than that for a CAS plant, but as the numbers of MBR plants increase, and as membrane costs fall, the life cycle cost differential will soon disappear, and the process advantages should lead to rapid uptake of the MBR system by the

j367

j 16 Wastewater Treatment by Membrane Bioreactors

368

wastewater-treatment industry. This chapter will therefore focus on membranebioreactor technology applied to municipal wastewater. 16.3 Membrane Bioreactors (MBR) 16.3.1 Membrane-Bioreactor Configurations 16.3.1.1 Membrane Materials and Options There is a large selection of commercial membranes that potentially can be used in MBR applications. Within the membrane industry, when classifying membranes, a distinction between polymeric membranes and inorganic membranes is made. Inorganic membranes are made either from metals or ceramics, the latter being the more common. Ceramic membranes are considered to be rather expensive and to date there is a very limited use of this material for wastewater treatment. As the costs of ceramic membranes are reduced and the membrane design is more geared towards wastewater applications one may foresee a gradual increase in the use of ceramic membrane-based systems. Polymeric membranes are by far the preferred material in wastewater treatment to date. In principle, most polymers can be used to manufacture membranes and there is a wide variety of commercially available polymeric membranes. In water and wastewater applications, however, most systems are based on a limited set of polymeric materials [12, 40, 55, 70–72]. The most common membrane materials are; polyvinyl diﬂuoride (PVDF), polyethylsulfone (PES), polyethylene (PE) and polypropylene (PP). Depending on the manufacturing technique applied, membranes made from these polymers can be produced with various geometries and speciﬁc physical properties. In this way they can be tailored to meet the speciﬁc demands of the application for which they are intended [12, 17, 55, 60, 90]. Membranes are generally formed as ﬂat sheets or with tubular/hollow-ﬁber geometry. With recent developments of manufacturing techniques, alternative products are also available on the market, for example multibore or multitube designs and self-supporting ﬂat sheets with channels. Given these variations membranes are commonly given the following classiﬁcations:

- FS – ﬂat sheet; - HF – hollow ﬁber; - CT – capillary tubular; - MT – multibore or multitubular. The membrane module and design will obviously depend on the type of membrane used. The ﬂat-sheet membranes are commonly constructed in a plate-andframe conﬁguration or as spiral-wound (SW) modules. HF/CT/MT membrane types are commonly manufactured into bundles that are installed in housing units or designed to be unconﬁned in the ﬂuid, that is, immersed units. The membranes are

16.3 Membrane Bioreactors (MBR) Table 16.1 Characteristics of membranes used in MBRs (Adapted from Ref. 24.)

Characteristics

Tubular membranes

Flat-sheet membranes

Hollow-fiber membranes

Arrangement Packing density Energy demand

External – recycling Low High (turbulent ﬂow)

External/submerged High Low

Cleaning

Efﬁcient þ physical cleaning possible Tubes or element

External/submerged Moderate Low–moderate (laminar ﬂow) Moderate

Replacement

Sheets

Backwashing possible Element

then constructed in such a way that they are self-supporting. In MBR applications the plate-and-frame FS and the HF/CT membrane modules are the preferred options (Table 16.1) [24]. A comprehensive presentation of all membrane types, modules and geometries is beyond the scope of this chapter, reference available membrane books for details [12, 17, 55, 60, 71, 77, 90]. The examples in Figure 16.2 are an illustration of a typical membrane module and installation. The most widespread FS membrane system is mounted as a spiral-wound (SW) unit. In the SW example the actual membrane module is shown together with how they are mounted inside a pressure vessel. A typical installation is shown where several pressure vessels are subsequently mounted in a stack. Pressurized HF units are typically operated as a crossﬂow system. In the example shown the HF modules are mounted vertically and arranged in a skid. Several variations of the theme can be found depending on the type of module and the manufacturer, where Figure 16.2 is not speciﬁc to a particular item.

Figure 16.2 Examples of membrane types (FS and HF) and typical installation/configuration.

j369

j 16 Wastewater Treatment by Membrane Bioreactors

370

Figure 16.2 (Continued).

In Table 16.2 some of the typical characteristics of the various types of membranes and conﬁgurations are given. In MBR systems SW membrane modules are not used as the channels within the spiral are prone to clogging when the feed water has high suspended-solids concentrations. Tubular membrane systems are not common either as they tend to become very expensive due to the low area to volume ratio. Commercial MBR systems today are normally based on immersed FS conﬁgurations or HF/CT conﬁgurations.

16.3 Membrane Bioreactors (MBR) Table 16.2 Typical characteristics of different types of membrane types and configurations.

Configuration

Area/volume ratio (m2/m3)

Cost

Advantages

Disadvantages

Plate & frame

400–600

High

Can be dismantled for cleaning

Complicated design

Spiral wound

800–1000

Low

Tubular

20–30

Very high

Low energy cost, Robust and compact Easily mechanically cleaned

Capillary

600–1200

Low

Hollow ﬁber

5000–40 000

Very low

Tolerates high solids (between tubular and hollow ﬁber) Can be backﬂushed

Cannot be backﬂushed Not easily cleaned Cannot be backﬂushed High capital and membrane replacement cost

Sensitive to pressure shocks

Compact design Tolerates high colloid levels

16.3.1.2 Process Configurations The process conﬁguration of a MBR plant will partly depend on the type of membrane used (FS or HF/CT) and partly on the design of the biological treatment. Figure 16.3 is an illustration of the typical conﬁgurations found in MBR treatment schemes. For immersed membrane designs the membrane modules are either inserted directly into the biological reactor or placed in a separate reactor constructed to hold the membrane modules only (schemes A and B in Figure 16.3). In a side-stream conﬁguration the membrane modules are placed outside the biological reactor and can be operated in deadend mode or in a crossﬂow mode with recycling of the concentrate stream back to the biological reactor (schemes C and D in Figure 16.3). In the immersed conﬁguration the treated water (permeate) is extracted from the membrane by vacuum (low pressure) in contrast to the side-stream option where the permeate is generally produced under pressure. The immersed systems are generally less energy intensive compared to the side-stream design where pumps are needed to maintain sufﬁcient crossﬂow velocities in the membrane unit or to overcome the increasing resistance due to build up of material on the membrane with deadend operation. To offset fouling and to generate crossﬂow conditions along the membrane, aeration is commonly used in the immersed systems. Each conﬁguration has its advantages and disadvantages where variations on the theme can be found from the different system manufacturers. These will be discussed in the following sections. The immersed MBR conﬁguration is generally the preferred option, particularly for medium- to large-scale municipal wastewater treatment plants [23, 29, 39, 40, 43, 45].

j371

j 16 Wastewater Treatment by Membrane Bioreactors

372

Figure 16.3 Typical configurations of MBR schemes, immersed vs. sidestream.

16.3.2 Membrane-Bioreactor Basics

The key elements, their interactions and impacts of these in the MBR process are shown in Figure 16.4. Ultimately, the feed characteristics of the wastewater to be treated will impact the design and operation of the process. The composition of the feed water as well as the required efﬂuent quality that the treated water needs to meet will also deﬁne the treatment scheme, particularly the conﬁguration of the biological process. In that the primary purpose of the membrane is to clarify the biologically treated water, the interaction between the biological process and the membrane process is the core of the technology. Due to the nature of the membrane ﬁltration it is evident that components in the feed water will be retained by the membrane and as this material is captured on the membrane it will cause fouling. One of the major drawbacks of MBRs is fouling, which is common for all membrane systems, where the efﬁcacy of the process is constrained by the accumulation of materials on the surface of or within the membrane resulting in a reduction in the membrane permeability. Several deﬁnitions of fouling can be found in the literature where both relative broad and explicit deﬁnitions are used. One deﬁnition deﬁnes fouling as a decline in time of ﬂux during operation when all operating parameters are kept constant [12]. This has further been revised to deﬁne fouling as a long-term phenomenon where irreversible fouling that builds up over time causes a ﬂux decline [78]. This deﬁnition has been further revised to distinguish between reversible and irreversible fouling as well as short-term and long-term fouling. As fouling is better understood other

16.3 Membrane Bioreactors (MBR)

Figure 16.4 Key elements and interactions of the MBR process.

reﬁnements of the deﬁnition are to be expected. Membrane fouling in MBR systems is caused by different substances and the mechanisms are rather complex and interrelated. Deposition of solids as a cake layer, pore plugging/clogging by colloidal particles, adsorption of soluble compounds and biofouling are some of the main forms of fouling that have been identiﬁed [21, 34, 40, 47, 56, 64]. Membrane fouling and strategies for fouling mitigation and amelioration are therefore the core of MBR operating parameters, system conﬁgurations and membrane-module designs. Aeration is a key element in the design and operation of MBRs with multiple objectives being included in the process. With respect to the biological component of the process, aeration is necessary to fulﬁll the oxygen demand of the aerobic degradation of compounds by the biomass. Aeration is also used in the membrane component of the process to prevent clogging of the membrane modules from solids concentrations and as a technique and means to prevent fouling of the membranes. The design of the aeration ports, intensities, air bubble characteristics and properties will therefore need to take into consideration the primary objectives of the aeration unit. A more detailed discussion of aeration is included in the sections on membrane fouling and deﬁning operating conditions. The hydraulics of the MBR process will impact both the biological process as well as the membrane process. From a biological point of view the hydraulics is signiﬁcant for the hydraulic retention time (HRT) that affects key operating parameters such as reactor volumes, mixed liquor suspended solids (MLSS), loading rates and so on. For the membrane process, hydraulics directly impacts the ﬁltration process and ultimately fouling developments. The design of the membrane module and operating

j373

j 16 Wastewater Treatment by Membrane Bioreactors

374

conditions are interlinked with ﬂow patterns and local hydrodynamic conditions on the membrane surface, which affect different fouling mechanisms. The hydraulics in the membrane system is also important when cleaning a fouled membrane needs to be done. A more detailed discussion of hydraulics in the membrane process is included in the sections on membrane fouling and how to deal with fouling. Overall, the MBR process needs to be designed and built with respect to two sets of operating requirements, those operating parameters needed to optimize the membrane component of the process and those necessary to achieve the desired biological conversions in the biological processes. The interactions and interdependence of the two key processes need to be understood in order to be able to develop treatment schemes that are efﬁcient both from a treatment perspective but also from a management and operational perspective. In the following sections some of the key elements in MBR systems will be discussed in more detail. 16.3.3 Membrane Fouling1) 16.3.3.1 Understanding Fouling Fouling is particularly a problem in AS-MBRs since the process deals with liquors having high concentrations of total solids as well as dissolved compounds such as extracellular polymeric substances (EPS). Fouling is deﬁned as reversible, that is, can be removed by backwashing strategies, or as irreversible, that is, fouling that is only recoverable by chemical cleaning, where the dominating fouling mechanism subsequently determines the performance of the process. Optimizing fouling control and cleaning strategies are therefore important aspects of developing and designing MBR processes. The complex nature of fouling in MBRs makes it difﬁcult to distinguish between which mechanisms or foulants are dominant and these may change with time during operation as well as due to variations in the feed characteristics, adjustments in operating parameters and so forth. Figure 16.5 gives an illustration of the main fouling mechanisms identiﬁed in membrane processes. Membrane fouling is manifested in various ways and certain types of fouling (reversible) can be removed by backwashing, that is, cake formation and loose depositions, while others are permanent (irreversible), fouling that is only recoverable by chemical cleaning [47]. A lot of effort and research has been done to gain a better knowledge of the phenomenon in the last 10–15 years. A literature review shows that in recent years the average number of articles found where the registered keywords are; fouling, membrane, colloids is around 25 and 40 [2]. Suspended solids are very often identiﬁed as a main foulant [6, 21] where the signiﬁcance of the submicrometer colloidal fraction in the suspended solids has been reported to correlate with membrane-fouling rates [47, 50, 69, 75, 83, 91]. In MBR processes fouling has also been attributed to extracellular polymeric substances (EPS) and 1) The basic aspects of fouling are discussed in Chapter 6, here practical aspects on fouling related to MBR are further considered.

16.3 Membrane Bioreactors (MBR)

Figure 16.5 Identification of potential causes for membrane fouling.

soluble microbial products (SMP) [56, 62, 69, 83]. Current research activities are very much focused on gaining a better understanding of the key fouling mechanisms, how they interact, and which factors inﬂuence their contribution to fouling as a whole. Many studies have distinguished between the relative contributions in per cent of the main fractions that cause fouling in MBR processes. An overview of several studies shows that there is a large variation when assigning fouling to the suspended solids fraction, the colloidal fraction or the soluble fraction [40]. Although the potential for cake formation and an increase of resistance to permeation by the layer formed is proportional to the suspended solids concentration, fouling by cakelayer formation is not commonly identiﬁed as the dominating mechanism in MBRs due to operation with relatively modest ﬂuxes. The soluble and colloidal fractions are assumed to be the dominant foulants. Several studies have reported the signiﬁcance of colloidal particles as an important factor contributing to fouling development. The contribution by colloids has been estimated to be responsible for anywhere between 20% to more than 60% of total measured fouling [8, 21, 35, 37, 38, 47, 69, 83]. The colloidal material may cause pore blockage and the deposition of these small particles may form very compact layers on the membrane surface. The remaining percentage of the total measured fouling is attributed to the soluble fraction. Studies of the soluble fraction have shown that the extracellular polymeric substances (EPS) are most critical with respect to fouling potentials of this fraction [67]. EPS are biological polymers of microbial origin that are predominantly made up of polysaccharides and proteins with small contributions of nucleic acids and lipids. EPS forms protective layers around the cells and also facilitates the interactions between the cells and the environment and are therefore essential for microbial survival. The EPS is often deﬁned as bound to the cell or soluble, which is the EPS found in the water phase due to breakup of ﬂocs or from cell lysis [26, 27]. The soluble portion is sometimes referred to as soluble microbial products (SMP), however, this fraction not only includes the EPS substances but can also include intermediates or end products from the biological conversion as well as endogenous cell decomposition. Studies have found correlations between EPS concentrations and membrane fouling in MBR systems. The polysaccharides and proteins have been shown to be the main foulants from EPS, however, contradicting reports are found as to whether it is the polysaccharides or proteins that have the greatest effect.

j375

j 16 Wastewater Treatment by Membrane Bioreactors

376

16.3.3.2 Dealing with Fouling MBR operation has aimed at trying to minimize fouling by maintaining operating parameters that give a sustainable process, that is, fouling over a long term with very low fouling rates. Within the industry, a sustainable ﬂux policy has been adapted, corresponding to a long-term ﬂux operation where fouling rates are below an economically accepted value. Many studies have been conducted to determine what a subcritical ﬂux operation really is. Fouling in MBR systems has recently been described as a three-stage process, each with a distinct pattern. The ﬁrst stage is characterized by a short-term fouling where the systems adjust to the set operating ﬂux and a kind of steady-state condition. This is followed by a second stage of longterm operation with relatively low fouling rates (expressed as an increase of TMP over time) until a sudden and sharp increase in TMP is observed (stage three). Several studies have reported the same observation regardless of the system studied [9, 13, 28, 64, 82]. This phenomenon has been explained by different theories, though it is not yet fully understood. However, it is apparent that deﬁning and operating MBR systems with a sustainable ﬂux, that is, subcritical ﬂux, is a key aspect and strategy for dealing with fouling (Figure 16.6) [13, 28]. The concept of critical ﬂux was introduced during the mid-1990s and deﬁned as a ﬂux below which fouling is absent or negligible. The basis of this concept is that for a

Figure 16.6 Long-term TMP profiles illustrating three fouling stages observed at subcritical flux.

16.3 Membrane Bioreactors (MBR)

given feed solution and operating condition in a crossﬂow mode, there is a ﬂux at which the transport of substances to the membrane exceeds the backtransport and removal of rejected substances and fouling of the membrane begins [5, 25, 84]. The concept gives a good understanding of membrane ﬁltration from a theoretical point of view but the concept has been prone to debate and various interpretations when complex systems such as MBR processes are considered. The basis of this concept, however, has been used to determine the optimal operating condition for complex systems as in MBRs, although the term critical ﬂux is used loosely here and should not be equated to the original concept. Coupling of the critical-ﬂux hypothesis with the process-cost optimization has led to the so-called sustainable ﬂux, which represents the operating ﬂux below which the fouling rate is economically acceptable for the plant operation. A stepping analysis approach has been proposed to determine at which ﬂux one could expect a sustainable operation. Figure 16.7 shows the stepping analysis proposed to determine the critical ﬂux (A) and an example of how this has been applied to a speciﬁc case (B) [28, 43]. It should be noted that the criticalﬂux value obtained is very speciﬁc for each case and is dependent on the nature and properties of the feedwater, the conﬁguration and operating conditions of the biological process, and the type of membrane modules used. The stepping analysis is, however, a tool one can use to determine the practical limitations of operation, that is, sustainable ﬂux, for given conditions and system speciﬁcations to achieve economical and efﬁcient operating parameters. Aeration is one of the most important parameters in the design and operation of MBR systems. Aeration is essential for the operation and design of the aerobic stage of the biological process with speciﬁc demands and needs expressed by the biological conversion. Aeration from this perspective is discussed in more detail in the section on biological operating conditions in MBRs. As indicated previously in Figure 16.4, aeration is used in the membrane process for air scouring and cleaning of the membrane module. Aeration in submerged MBR, particularly for hollow ﬁber systems, also induces a lateral movement that generates a shear force on the membrane from the surrounding liquid. The overall effect is a function of the aeration intensity and how much movement is achieved [16, 23].

Figure 16.7 Determination of critical flux in MBRs; (a) method, (b) applied.

j377

j 16 Wastewater Treatment by Membrane Bioreactors

378

16.3.3.3 Cleaning Fouled Membranes Membrane cleaning is often deﬁned by the type of cleaning and the frequency of cleaning. The use of air scouring, periodic backwashing or back pulsing during operation are generally considered to be fouling-mitigation strategies to reduce the effect of short-term fouling. Membrane cleaning as an operational term is often associated with a more extensive action that needs to be undertaken when fouling of the membrane reaches a point where production of treated water is reduced to a critical level. Cleaning protocols are deﬁned by applying both physical and chemical techniques and a distinction between maintenance and recovery cleaning is also made. Maintenance cleaning is commonly used to reduce long-term fouling development and subsequently to increase the frequency between intensive cleaning necessary for recovery of permeability. Maintenance cleaning is generally performed in situ with the membranes kept in place (CIP). From an operation point of view, maintenance cleaning is a procedure that is done at regular intervals (3–7 days) where cleaning agents are applied at low concentrations. The frequency is determined by the feed characteristics and operating conditions, while the procedure is a manufacturer speciﬁc recommendation. For recovery cleaning the membrane is commonly removed from the process line and higher concentrations of cleaning reagents are used. Several cleaning steps are employed and can last for several hours. As this directly impacts the production capacity of the plant and operation and maintenance routines, systems are designed to try and reduce recovery cleaning frequencies between 1–3 times per year. Physical cleaning can be performed by generating high shear forces to remove any deposits on the membrane surface, altering the hydrodynamics around the membrane, air scouring, movement (i.e., vibrations, ultrasound). Mechanical cleaning can also be induced to remove deposits depending on membrane geometry, membrane material and operating criteria. Although physical cleaning may quite effectively remove cake layers or deposited solids on the membrane surface it needs to be supplemented with chemical cleaning to fully restore the membrane permeability. Organic and inorganic fouling caused by adsorption or deposition can only be removed through chemical cleaning. In principle, three strategies are employed; oxidation of organic compounds on the membrane (commonly targeting biofouling constituents) coupled with caustic solutions to dissolve and remove organic compounds, and acidic solutions to remove inorganic/mineral compounds. In practice, cleaning protocols are generally deﬁned and recommended by the manufacturers and suppliers of MBR systems. With respect to recovery cleaning the choice of cleaning agent and steps is determined by whether the aim is to remove organic or inorganic foulants. A survey of cleaning protocols shows that the chemical products recommended and their concentrations are quite similar for all the systems available on the market. The backbone of the protocols include; applying hypochlorite to remove biofouling (typical concentration range 200–2000 ppm), increasing pH with sodium hydroxide (typical concentration range 150–4000 ppm) to dissolve organic compounds, and decreasing pH with acids, commonly citric acid and oxalic acid (typical concentration range up to 5000 ppm) to dissolve inorganic compounds. The

16.3 Membrane Bioreactors (MBR)

choice of chemicals and cleaning steps are very much dependent on the system supplier. There are a limited number of studies in the literature focusing on MBR cleaning protocols. In summary, the following generalized observations can be made. The membrane-cleaning protocols recommended are commonly based on the perception of what the dominant fouling mechanism or foulant has been. The overall cleaning processes generally involve many steps/stages and are often time consuming, a fact that affects operation time that may cause a loss in overall production of permeate. A combination of physical and chemical cleaning methods appears to provide the best permeability-recovery efﬁciency. Recoveries well over 90% and close to 100% are commonly reported. One challenge, however, is the lack of standard criteria to deﬁne membrane-cleaning parameters and measurements of ﬂux/permeability recovery, and how these may ultimately affect membrane properties and membrane lifetime as a function of cleaning frequency. Finally, there are very limited references on waste management and handling of spent cleaning agents or the removed substances from the cleaning action [49, 51, 61, 74, 86, 87]. 16.3.4 Defining Operating Conditions and Parameters in MBR Processes 16.3.4.1 Biological Operating Conditions Biological treatment of wastewater utilizes the conversion of organic and inorganic matter into products by micro-organisms that are either easy to remove from the water (i.e., biomass growth) or converted to nonharmful substances. The biological conversions are a function of the biological community present in the process and the conditions necessary for the existence of the community, for example aerobic vs. anaerobic degradation. The biological process much depends on the treatment scheme and the target compounds to be removed by biological degradation [53]. A comprehensive assessment of the biological processes is beyond the scope of this chapter, there are many dedicated books available on the subject. There are, however, some speciﬁc details on the impact of coupling the biological process with the membrane process (Figure 16.4) that need to be highlighted with respect to implications on the operation of MBR processes. Compared to conventional activated-sludge process, the biomass that is formed in the MBR system differs in composition and characteristics. A ﬁltration test of sludge from CAS was found to give irreversible fouling compared to MBR sludge that gave reversible fouling [20]. One of the characteristics of the AS-MBR process is the concentration of biomass that inﬂuences the rheological properties of the sludge. During the development of MBR systems, very high sludge concentrations have been reported, however, from an energy-efﬁcient perspective the recommended concentration has been stated to be between 10–15 g/L MLSS [1, 30]. Under these conditions the viscosity of the sludge was found to differ from CAS sludge and in general the viscosity was found to increase with increasing MLSS. The oxygen demand is dependent on the biological process with respect to whether it is designed to meet secondary efﬂuent standards or not. In most cases MBR

j379

j 16 Wastewater Treatment by Membrane Bioreactors

380

systems are chosen due to their capability to produce a high-quality efﬂuent and as such are designed to include nutrient removal. In this case the oxygen necessary to maintain a micro-organism community that degrades both organic matter and converts ammonium to nitrite/nitrate is required. The reader should refer to the appropriate literature on biological treatment for detailed explanations on how to calculate the oxygen demand for a given biological process. Aeration in MBR systems is a key issue where mass transfer of oxygen to the system can be calculated using the classical kLa equation, expressed as oxygen transfer rate (OTR); OTR ¼ akL aðbDO DOÞ where kLa is the overall mass-transfer coefﬁcient and DO and DO are the saturated and dissolved oxygen concentrations. a and b are correction factors for the mass transfer rates and saturation concentrations commonly determined for clean water and therefore compensate for application to wastewater. Studies have shown that a decrease in OTR can be observed with increasing solids concentration where an exponential relationship between the a factor and MLSS concentration has been reported [30, 40]. The main impact has been reported to relate to bubble behavior in the liquid where higher concentrations appear to promote coalescence of bubbles and thus a reduction in the interfacial area expressed in the kLa term. Studies on bubble aeration have shown that a greater resistance to oxygen mass transfer is observed with increasing viscosity and correlations between the a-factor and viscosity have been proposed. a ¼ mx where m is the viscosity (kg/(m s)) and x is the correlation exponential [40]. An effect of the increase in MLSS concentration in MBRs is a decrease in the a factor, where measurements at a MLSS of 12 g/L MLSS gave an a value of 0.6 compared to values of 0.8 typical of CAS sludge at concentrations of 3–5 g/L MLSS [15, 30]. The practical implication of this is that higher aeration rates and intensities are necessary in MBR systems compared to CAS systems. In principle, the biological conversions in the MBR are performed according to what happens in the CAS process and as such the conversions achieved are not substantially different. The difference in the sludge properties do, however, affect the operational parameters related to pumping/circulation of the sludge and aeration of the sludge to maintain the necessary dissolved oxygen for the aerobic stage. Studies have also indicated that the response of the biomass to dissolved oxygen concentrations may promote or increase fouling potentials due to changes and stresses in the conversion mechanisms of the biological community. This stress has been shown to affect EPS production and composition of the EPS, one of the main foulants identiﬁed in MBR processes. Operating the biological process at optimal conditions to maximize the desired conversions while minimizing production or generation of potential foulants is therefore a key issue in the sustainable operation of MBR processes. Taking into account that aeration is an energy-demanding component and represents around 40% of the energy consumption in MBRs, aeration for the biological process in MBRs is a key operating parameter.

16.3 Membrane Bioreactors (MBR)

16.3.4.2 Membrane Filtration Operation Membrane operation will much depend on the system conﬁguration or mode of operation; deadend, crossﬂow or immersed. Although the schemes may differ, the underlying objective is to reduce fouling by the way the membrane unit is operated [11, 19, 22, 46, 52, 63, 68, 80]. Identifying and operating the system with a sustainable ﬂux is a common theme. From a practical point of view the stepping analysis is a useful tool to identify a subcritical zone of operation where it is possible to achieve long-term and reasonably stable operation of the system [32, 43, 64]. Determining the subcritical zone needs to be done experimentally for each speciﬁc condition. When the critical-ﬂux is found it is possible to identify the subcritical working area and the overall design of the MBR plant to account for daily variations in ﬂow. An example of expressing this area of operation is illustrated in Figure 16.8. From the graph it is apparent that fouling can be kept at a minimum (expressed as an increase in TMP with increase in permeability) as long as one keeps in the subcritical zone. Membrane aeration is a key operational parameter in that air scouring is used to keep solids from the membrane surface and to reduce fouling. Aeration is very energy demanding and the design and operation of systems in MBRs is where a lot of focus has been made in recent years to reduce this aspect of MBR operation. A key parameter for design and operation of MBR systems is therefore the speciﬁc aeration demand (SAD). A challenge within the industry is to deﬁne this parameter in such a way that different systems can be compared in a realistic manor. SAD has been normalized with respect to membrane surface area or to permeate volume produced [40]. A number of studies have demonstrated that ﬂux increases linearly with increasing aeration rates until a threshold above which no further effect of increasing the aeration intensity can be observed [36, 44, 76]. Indeed, increasing the aeration beyond this threshold may have a negative impact on the performance of the membrane ﬁltration

Figure 16.8 Experimental determination of subcritical zone of operation [64].

j381

j 16 Wastewater Treatment by Membrane Bioreactors

382

unit. Too high shear forces caused by the high aeration intensity may break up the ﬂocs in the suspended solids, increasing the colloidal fraction that in turn increases fouling potential. Reports have also been published that show how more intensive aeration not only damages ﬂoc structures but also can release EPS-based foulants bound in the ﬂoc structure [36, 37, 40]. However, references in the literature that report correlations between aeration rates and effects on changes in colloidal-particle characteristics as a consequence of membrane aeration cannot be found. The threshold effect and the impact aeration intensity has on an increase in the colloidal fraction expressed as an increase in the differential number percentage of the submicrometer colloidal fraction is illustrated in Figure 16.9. The increase in this fraction also correlated with higher

Figure 16.9 Effect of aeration intensity on membrane filtration performance [36].

16.3 Membrane Bioreactors (MBR)

fouling rates. Membrane ﬁltration operation has also been improved by implementing alternative aeration strategies, mainly with the focus to reduce the aeration energy demand and costs. Intermittent aeration strategies and ﬂuctuations in aeration intensities over periods are approaches that MBRs suppliers have investigated. The objective is to reduce overall energy requirements for aeration while maximizing the fouling mitigation effects of aeration. The speciﬁcs of these operating parameters are very much system dependent and each supplier has their own preferred emphasis and recommendations. Removing or reducing the submicrometer particulate fraction is one approach to improving the membrane ﬁltration stage. Coagulation and ﬂocculation have been attempted as a means to improve the performance and thus gain a better understanding of dominating fouling mechanisms. Coagulation may also be used as a technique to reduce the dissolved organic matter content (particularly the EPS) and is also offered by the industry as a means to get a more stable operation of the membrane unit [14, 41, 65, 66, 81]. Other techniques such as periodic backwashing or relaxation techniques are also now standard modes of operation for most MBR suppliers. Backwashing the membranes by reversing the ﬂow can generate enough ﬂux to lift off deposits on the membrane and remove cake-layer fouling. FS membrane modules are not designed to be backﬂushed and backwashing is there predominantly applied in HF/CT systems. Backwashing in MBRs is synonymous with backwashing in ﬁltration systems in general. Introducing backwashing makes the MBR a discontinuous process where the frequency and backwash volume ultimately affect the overall production of treated water. Backwashing adds an operational dimension to the process but reducing fouling and increasing the time between major cleaning by far outweigh the disadvantages of introducing a backwashing mode of operation. Backwashing is further enhanced when it is coupled with air scouring [31]. An alternative strategy to backwashing is relaxation, which can also be applied to the FS systems. This techniques is based on stopping the production of permeate, that is, applying vacuum, and allowing the system to rest for a short period. When air scouring is applied during relaxation the shear forces generated by the air bubbles can more efﬁciently remove the deposits accumulated on the membrane surface during production. The intermittent operation and frequency of relaxation is system dependent and determined by the MBR supplier. Lastly, maintenance cleaning as described in the section above on cleaning fouled membranes is also implemented as part of the normal operating parameters for commercial MBR systems. 16.3.4.3 Optimizing MBR Operations When ﬁrst commercialized, MBR processes were considered to be very expensive systems and only suitable for small-scale plants and for very speciﬁc applications. Capital costs have dropped drastically with the development of several commercially available systems and the treatment scheme is competitive even for large treatment plants. The operation and maintenance costs have gone through an evolution. In the infancy of the technology a major cost item was theanticipated membrane-replacement

j383

j 16 Wastewater Treatment by Membrane Bioreactors

384

Figure 16.10 Illustration of typical energy demands in the operation of a MBR process.

costs, now this item has dropped signiﬁcantly due to better and cheaper production of membrane modules as well as an increase in lifetime expectancy gained from operating experiences. Today, the energy demands are by far the largest operating costs, in particular the need for aeration both for the biological process and of the membrane process (Figure 16.10). As the aeration of the membrane unit is closely linked with fouling control and mitigation, continued studies to understand the complex nature of membrane fouling in MBR systems as well as the design and operation of membrane modules will be a major activity in the development of MBR systems in the future.

16.4 Prospects and Predictions of the MBR Process 16.4.1 Developments and Market Trends

Attempts at predicting the future needs and investments in the water and sanitation sector have been done, however, by comparing the various regions in the world and making assessments both of urban and rural needs is not easy. Some studies though have indicated the extent and challenges for the water industry in general. Investment in water quality in developing nations will help drive an estimated 5.9% annual increase (including price increases) in demand for water-treatment products through 2009, according to a new study, World Water Treatment, by research ﬁrm The Freedonia Group, Inc. The same ﬁrm has also recently made an assessment of the advancement of membrane technology in a study entitled Membrane Separation Technologies. This study stipulated a yearly increase of 7.8% in the demand for membrane materials with a total value of membrane systems (including equipment such as pumps and piping) reaching USD 4.8 billion in 2004. Water and wastewater treatment has been identiﬁed as the largest end use for membrane

16.4 Prospects and Predictions of the MBR Process

Figure 16.11 Stipulated growth in the water and wastewater industry to 2015 (Adapted from Ref. 59.)

materials, where they will continue to dominate. Implementation and expansion in consumer applications combined with replacement sales to municipal and industrial customers are suggested as the main reasons for this. The global water market has been estimated at a total value of around 224 billion D with an anticipated annual growth of around 16–20% depending on the market segment (Figure 16.11). Drinking-water production is stipulated to have the largest growth with a doubling of the market value in the period 2000–2015. The market for wastewater treatment is the largest sector with an anticipated growth of around 43% for the same period. Membrane technology will of course play an important role in this market. Cross-ﬂow membrane systems are expected to grow from 4.8 billion D in 2004 to 6.5 billion D in 2007 on a global basis where desalination has been reported to represent about 1/3 of this growth. The fastest growing segment, however, has been predicted to be the development of membrane-bioreactor systems for wastewater treatment with a yearly growth estimated at 15% [42, 57–59]. Looking at the water and sanitation sector in general, certain trends may be found regarding the advancement and implementation of membrane technology in environmental engineering. Microﬁltration membranes (MF) are recognized as accounting for the largest portion of the market. The nature of MF separation make them by far the most widely used membrane process where they can either constitute the ﬁnal separation stage or are applied as pretreatment options in, for example, reverseosmosis (RO) systems. RO for desalination of brackish water or seawater to produce potable water is a well-established industry. The cost of desalination is expected to drop drastically (presented at Fourth World Water Forum, Mexico, 2006) and reports about the implementation of large desalination plants to produce fresh water can often be found in the news media. In areas of the world experiencing water shortages or high demands on limited fresh-water resources, wastewater recycling and reuse is becoming a necessity. Membrane technology is a central and key element in implementing sustainable solutions for wastewater recycling and reuse. Numerous examples of membrane systems success for this application can be found and there is an increasing interest in implementing the available technology.

j385

j 16 Wastewater Treatment by Membrane Bioreactors

386

Secondary treatment of wastewater is a large and energy-intensive process, generally based on the conventional activated-sludge process. The MBR process by itself can fulﬁll secondary treatment on its own – doing it much better and in less space. It is rapidly becoming accepted as the best available technology (BAT). Since commercial development of the MBR process in the late 1980s a growing reference list of installations and plant can be found as well an increasing number of manufacturers and suppliers. The list of references keeps increasing and there are currently around 3000 MBR plants in operation or under construction worldwide [48, 73, 89]. The MBR market is dominated by the earliest developers Zenon in Canada (now GE Water technologies) and Kubota in Japan, closely followed by Wehrle Werk in Germany. USFilter (now part of Siemens) and Mitsubishi Rayon have also emerged as major suppliers. There are now around 30 MBR suppliers worldwide though the markets are dominate by the larger suppliers, that is, Zenon-GE and Kubota representing 63 and 30% of the European market, respectively [48]. The growth and trends of MBR installations in the European market and the main suppliers to date are illustrated in Figure 16.12. Similarly, the situation in North America and an overview of the registered installations worldwide for some of the major suppliers is illustrated in Figure 16.13. The ﬁrst MBR installations were for relatively small treatment plants, the general impression at the time, that MBR systems were only suitable for small-scale installations. The ﬁrst full-scale MBR plants were designed to treat wastewater for around 3–4000 person equivalents (i.e., the Porlock plant, UK, commissioned in 1998), however, treatment plants commissioned in the late 1990s/early 2000s showed a steady increase in size and capacity (i.e., the Nordkanal plant, Germany, commissioned in 2004 for 80 000 p.e., max. 48 000 m3/day). The largest MBR plant announced so far is the Brightwater plant, USA, which will have a capacity of 495 000 m3/day when it is commissioned in 2010/2011 [48, 73]. Given these trends it is clear that MBR technology will be a central and important option for advanced wastewater treatment in the future. 16.4.2 An Overview of Commercially Available Systems

The commercially available MBR systems can be classiﬁed into two distinct groups; those based on ﬂat-sheet membranes and those using tubular or hollow-ﬁber membranes. In literature reviews one can ﬁnd references to other membrane geometries being investigated such as multibore tubular membranes or self-supporting membrane sheets with integrated canals. There are no full-scale installations with these kinds of membranes to date. With the dynamic trend of the market, new suppliers as well as new products and novel solutions are to be expected. Given this situation, it is beyond the scope of this chapter to give a comprehensive and complete presentation of all MBR suppliers and products. In the following sections a brief overview will be given of the main systems being used, represented by the main suppliers, and the examples given are only intended to give newcomers to the technology a better understanding of the industry.

16.4 Prospects and Predictions of the MBR Process

Figure 16.12 Development of the European MBR market (Adapted from Ref. 48.)

Figure 16.13 MBR installations in North America and worldwide (Adapted from Ref. 89.)

j387

j 16 Wastewater Treatment by Membrane Bioreactors

388

16.4.2.1 Flat-Sheet MBR Designs and Options The ﬂat-sheet MBR module, originally developed by Kubota, consists of two rectangular sheets of membrane that are wrapped around and welded onto a panel to make a cartridge. The support panel has a core structured for spacer material to allow water to move freely within the cartridge. These membrane cartridges are subsequently hung vertically with a deﬁned space between each cartridge as speciﬁed by the various suppliers. Each cartridge is ﬁtted with a nozzle and connected by tubes to a collection pipe from which the permeate is extracted by vacuum. The mode of ﬂow is from the outside and into each cartridge. The spacing between each cartridge is optimized to allow for bubbles generated during air scouring preventing accumulation of solids between the panels causing clogging and fouling of the membrane. A view of the membrane cartridge and how they are arranged in the reactor are illustrated in Figure 16.14. An alternative design to the membrane cartridges described above is the rotating disk unit. These membrane modules are constructed on a similar principle but are mounted on discs instead of panels. The discs are then mounted on axles enabling the stack of discs to be rotated. The permeate is extracted either through the shaft by vacuum (the Hitachi Plant design) or tubes mounted on each disc (the Huber Technology design), see Figure 16.15. One of the advantages claimed by the manufacturers of this design is that the rotation can add to generating turbulent ﬂow regimes that help reduce membrane fouling and potentially increases the rate of permeate production. 16.4.2.2 Tubular/Hollow-Fiber MBR Designs and Options For the tubular or hollow-ﬁber MBR systems the submerged or immersed conﬁguration is generally the preferred option. There are several suppliers offering such systems but the concept is essentially the same. Membranes are put together in modules (rectangular or circular units) that are then mounted into cassettes. The cassettes are the building blocks of the system, forming the modular design typical of MBR systems. The capacity of the treatment plant is thus a function of how many modules are necessary to treat the design ﬂow. As air scouring is an essential part of

Figure 16.14 Illustration of a FS MBR configuration: FS cartridges, and arrangement in reactor.

16.4 Prospects and Predictions of the MBR Process

Figure 16.15 Examples of the rotating-disc membrane reactor design.

the system, the aeration devices are normally integrated into the module design. Examples of some of the available solutions are illustrated in Figure 16.16. Examples of side-stream conﬁgurations are shown in Figure 16.17. These solutions are generally pressurized systems, though the transmembrane pressure or feed pressure is less than in conventional crossﬂow membrane systems. The mode of operation may be deadend or with recirculation of the concentrate. On side-stream systems a feed pump is commonly used to pressurize the membrane modules and to create the circulation of the feed water, where this applies. Some systems apply airlift principles to generate a crossﬂow mode of operation, typically in an inside-out conﬁguration using tubular membranes. An added beneﬁt of airlift systems is the effect of dual-media ﬂow where bubbles inside the tubes help generate high shear forces that decrease fouling of the membranes. Another claimed beneﬁt of sidestream conﬁgurations is from an operation and maintenance perspective where cleaning or replacing membrane modules is easier and more practical due to the accessibility. Given the history and development of environmental engineering, with wastewater treatment in particular, as it is understood today, the membrane bioreactor is quite a recent invention. Commercialization and implementation of the technology has only been going on in the last couple of decades, and so, not surprisingly, MBR technology is still in a period of intense development [33]. The current MBR business

j389

j 16 Wastewater Treatment by Membrane Bioreactors

390

Figure 16.16 Examples of immersed tubular/hollow-fiber membrane systems.

can be found among the wastewater-equipment suppliers, either the specialist builders of wastewater treatment equipment who have acquired membrane technology (possibly by buying a membrane specialist company), or by membrane system manufacturers who have developed an expertise in wastewater treatment. Irrespective of point of entry to the business, as the technology gains recognition as the best available technology (BAT) for wastewater treatment, the number of both new suppliers and new systems is expected to increase. In the foreseeable future, three key areas of system investigation can be identiﬁed independent of supplier and solution: the nature of the membrane coupled with operating energy consumption, air/gas handling and the bioreaction itself.

Figure 16.17 Examples of side-stream systems based on airlift or deadend mode.

References

Nomenclature

AS CAS RO MBR AS-MBR FS HF MF UF NF EPS TMP PVDF PES PE PP DO

Activated sludge Conventional activated sludge Reverse osmosis Membrane bioreactor Activated-sludge membrane bioreactor ﬂat sheet Hollow ﬁber Microﬁltration Ultraﬁltration Nanoﬁltration Extracellular polymeric substances Transmembrane pressure Polyvinyl diﬂuoride Polyethylsulfone Polyethylene Polypropylene Dissolved oxygen (mg/L)

References 1 Adham, S., Gagliardo, P., Boulos, L., Oppenheimer, J. and Trussel, R. (2001) Feasibility of the membrane bioreactor process for water reclamation. Water Science and Technology, 43 (10), 203–209. 2 Aimar, P. Recent Progress in Understanding Particle Fouling of Filtration Membranes, Laboratoire de Genie Chimique – CNRS – Universite Paul Sabatier, http:// www.membrane.unsw.edu.au/. 3 Arden, E. and Lockett, W.T. (1914) The oxidation of sewage without the aid of ﬁlter. Part I and II. Journal of the Society of Chemical Industry, 33 (10), 523–539, 1122–1124. 4 Alonso, E., Solis, G.J. and Riesco, P. (2001) On the feasibility of urban wastewater tertiary treatment by membranes: a comparative assessment. Desalination, 141, 39–51. 5 Bacchin, P. (2004) A possible link between critical and limiting ﬂux for colloidal systems: consideration of critical deposit

6

7

8

9

formation along a membrane. Journal of Membrane Science, 228, 237–241. Bae, T.H. and Tak, M.T. (2005) Interpretation of fouling characteristics of ultraﬁltration membranes during the ﬁltration of membrane bioreactor mixed liquor. Journal of Membrane Science, 264 (1–2), 151–160. Baker, M.N. (1981) The Quest for Pure Water, vol. 1, 2nd edn, American Water Works Association, Denver. Bouhabila, E.H., Ben Aim, R. and Buisson, H. (1998) Microﬁltration of activated sludge using submerged membrane with air bubbling (application to wastewater treatment). Conference on Membranes in Drinking and Industrial Water Production, Amsterdam. Brookes, A., Jefferson, B., Guglielmi, G. and Judd, S.J. (2006) Sustainable ﬂux fouling in a membrane bioreactor: Impact of ﬂux and MLSS. Separation Science and Technology, 41 (7), 1279–1291.

j391

j 16 Wastewater Treatment by Membrane Bioreactors

392

10 Carmen, C., Teodosiu, M., Kennedy, M.D., Van Straten, H.A. and Schippers, J.C. (1999) Evaluation of secondary reﬁnery efﬂuent treatment using ultraﬁltration membranes. Water Resources, 33 (9), 2172–2180. 11 Chang, S. and Fane, A.G. (2000) Filtration of biomass with axial inter-ﬁbre upward slug ﬂow: performance and mechanisms. Journal of Membrane Science, 180, 57–68. 12 Cheryan, M. (1998) Ultraﬁltration and Microﬁltration. Handbook, Technomic Publishing Company, Inc., USA. 13 Cho, B.D. and Fane, A.G. (2002) Fouling transients in nominally cub-critical ﬂux operation of a membrane bioreactor. Journal of Membrane Science, 209 (2), 391–403. 14 Choksuchart, P., Heran, M. and Grasmick, A. (2002) Ultraﬁltration enhanced by coagulation in an immersed membrane system. Desalination, 145, 265–272. 15 Cornel, P., Wagner, M. and Krause, S. (2003) Investigation of oxygen transfer rates in full scale membrane bioreactors. Water Science & Technology, 47 (11), 313–319. 16 Cote, P., Buisson, H. and Praderie, M. (1998) Immersed membranes activated sludge process applied to the treatment of municipal wastewater. Water Science and Technology, 38 (4–5), 437–442. 17 Crespo, J.G. and B€odekker, K.W. (eds) (1994) Membrane Processes in Separation and Puriﬁcation, Kluwer Academic Publishers, Dordrecht. 18 Davies, W.J., Le, M.S. and Heath, C.R. (1998) Intensiﬁed activated sludge process with submerged membrane microﬁltration. Water Science and Technology, 38 (4–5), 421–428. 19 Defrance, L. and Jaffrin, M.Y. (1999) Comparison between ﬁltrations at ﬁxed transmembrane pressure and ﬁxed permeate ﬂux: application to a membrane bioreactor used for wastewater treatment. Journal of Membrane Science, 152, 203–210. 20 Defrance, L. and Jaffrin, M.Y. (1999) Reversibility of fouling formed in activated

21

22

23

24

25

26

27

28

29

30

sludge ﬁltration. Journal of Membrane Science, 157, 73–84. Defrance, L., Jaffrin, M.Y., Gupta, B., Paullier, P. and Geaugey, V. (2000) Contributions of various constituents of activated sludge to membrane bioreactor fouling. Bioresource Technology, 73, 105–112. Duin, O., Wessels, P., van der Roest, H., Uijterlinde, C. and Schoonewilde, H. (2000) Direct nanoﬁltration or ultraﬁltration of WTTP efﬂuent? Desalination, 132, 65–72. Fane, A.G., Chang, S. and Chardon, E. (2002) Submerged hollow ﬁbre membrane module - design options and operational considerations. Desalination, 146, 231–236. Fane, A. (2002) Membrane bioreactors: design and operational options. Filtration & Separation, 39 (5), 26–29. Field, R.W., Wu, D., Howell, J.A. and Gupta, B.B. (1995) Critical ﬂux concept for microﬁltration fouling. Journal of Membrane Science, 100, 259–272. Flemming, H.C., Schaule, G., Griebe, T., Schmitt, J. and Tamachkiarowa, A. (1997) Biofouling - the Achilles heel of membrane processes. Desalination, 113, 215–225. Flemming, H.C. and Wingender, J. (2001) Relevance of microbial extracellular polymeric substances (EPSS) - part 1: Structural and ecological aspects. Water Science and Technology, 43, 1–8. Guglielmi, G., Chiarani, D., Judd, S.J. and Andreottola, G. (2007) Flux criticality and sustainability in a hollow ﬁbre membrane bioreactor for municipal wastewater treatment. Journal of Membrane Science, 289, 241–248. G€ under, B. and Krauth, K. (1998) Replacement of secondary clariﬁcation by membrane separation – results with plate and hollow ﬁber modules. Water Science and Technology, 38 (4–5), 383–393. G€ under, B. and Krauth, K. (2000) Excess sludge production and oxygen transfer in MBR. Proceedings ATSV conference, 8–9 February.

References 31 Hillis, P., Padley, M.B., Powell, N.I. and Gallagher, P.M. (1998) Effects of backwash conditions on out-to-in membrane microﬁltration. Desalination, 118, 197–204. 32 Howell, J.A. (1995) Subcritical ﬂux operation of microﬁltration. Journal of Membrane Science, 107, 165–171. 33 Howell, J.A. (2002) Future research and developments in the membrane ﬁeld. Desalination, 144, 127–131. 34 Huang, L. and Morissey, M. (1998) Fouling of membranes during microﬁltration of surimi wash water: Roles of pore blocking and surface cake formation. Journal of Membrane Science, 144, 113–123. 35 Itonaga, T., Kimura, K. and Watanabe, Y. (2004) Inﬂuence of suspension viscosity and colloidal particles on permeability of membrane used in membrane bioreactor (MBR). Water Science and Technology, 50, 301–309. 36 Ivanovic, I. and Leiknes, T. Impact of aeration rates on particle colloidal fraction in the bioﬁlm membrane bioreactor (BFMBR). Proceedings, IWA 4th International Membrane Technologies Conference, 15–17 May 2007, Harrogate, UK. 37 Ji, L. and Zhou, J. (2006) Inﬂuence of aeration on microbial polymers and membrane fouling in submerged membrane bioreactors. Journal of Membrane Science, 276, 168–177. 38 Juang, L.C., Tseng, D.H. and Yin Lin, Y.H. (2007) Membrane processes for water reuse from the efﬂuent of industrial park wastewater treatment plant: a study on ﬂux and fouling of membrane. Desalination, 202, 302–309. 39 Judd, S. (2002) Submerged membrane bioreactors: plate or hollow ﬁbre? Filtration þ Separation, 39, 30–31. 40 Judd, S. (2006) The MBR Book, Elsevier, Amsterdam. 41 Kim, J.-S., Akeprathumchai, S. and Wickramasinghe, S.R. (2001) Flocculation to enhance microﬁltration. Journal of Membrane Science, 182, 161–172.

42 Laine, J.-M., Vial, D. and Moulat, P. (2000) Status after 10 years of operation - overview of UF technology today. Desalination, 13 (1), 17–25. 43 Le-Clech, P., Jefferson, B., Chang, I.-S. and Judd, S.J. (2003) Critical ﬂux determination by the ﬂux-step method in a submerged membrane bioreactor. Journal of Membrane Science, 227, 81–93. 44 Le-Clech, P., Jefferson, B. and Judd, S.J. (2003) Impact of aeration, solids concentration and membrane characteristics on the hydraulic performance of a membrane bioreactor. Journal of Membrane Science, 218, 117–129. 45 Le-Clech, P., Jefferson, B. and Judd, S.J. (2005) Comparison of submerged and side-stream tubular membrane bioreactor conﬁgurations. Desalination, 173, 113–122. 46 Le-Clech, P., Fane, A.G., Leslie, G. and Childress, A. (2005) The operators perspective. Filtration þ Separation, 42, 20–21. 47 Leiknes, T. and Ødegaard, H. (2007) The development of a bioﬁlm membrane bioreactor. Desalination, 202, 135–143. 48 Lesjean, B. and Huisjes, E.H. Survey of European MBR market, trends and perspectives. Proceedings, IWA 4th International Membrane Technologies Conference, 15–17 May 2007, Harrogate, UK. 49 Liao, B.Q., Catalan, L.J.J., Droppo, I.G. and Liss, S.N. (2004) Impact of chemical oxidation on sludge properties and membrane ﬂux in membrane separation bioreactors. Journal of Chemical Technology and Biotechnology, 79, 1342–1348. 50 Li, M., Xiufen, L., Du, G., Chen, J. and Shen, Z. (2005) Inﬂuence of the ﬁltration modes on colloid adsorption on the membrane in submerged membrane bioreactor. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 264 (1–3), 120–125. 51 Lim, A.L. and Bai, R. (2003) Membrane fouling and cleaning in microﬁltration of

j393

j 16 Wastewater Treatment by Membrane Bioreactors

394

52

53

54

55

56

57

58

59

60

61

62

63

activated sludge wastewater. Journal of Membrane Science, 216, 279–290. Lodge, B., Judd, S.J. and Smith, A.J. (2004) Characterisation of dead end ultraﬁltration of biotreated domestic wastewater. Journal of Membrane Science, 231, 91–98. Mallevialle, J., Odendaal, P.E. and Weisner, M.R. (1996) Water Treatment Membrane Processes, McGraw-Hill, New York. Metcalf & Eddy (2003) Wastewater Engineering: Treatment and Reuse, 4th International edn, McGraw-Hill, New York. Mulder, M. (1997) Basic Principles of Membrane Technology, Kluwer Academic Publishers, The Netherlands. Nagaoka, H., Ueda, S. and Miya, A. (1996) Inﬂuence of bacterial extracellular polymers on the membrane separation activated sludge process. Water Science Technology, 34 (9), 165–172. News Release, December 2000, RO, UF, MF World Markets, www.macilvainecompany. com, published by the McIlvaine Company. News Release, July 2002, RO, UF, MF World Markets, www.macilvainecompany.com, published by the McIlvaine Company. News Release, 2004, www. macilvainecompany.com, published by the McIlvaine Company. Noble, R.D. and Stern, S.A. (1995) Membrane Separations Technology. Principles and Applications, Membrane Sciences and Technology Series, vol. 2, Elsevier. Nuengjamnong, C., Cho, J., Polprasert, C. and Ahn, K.H. (2006) Extracellular polymeric substancess inﬂuence on membrane fouling and cleaning during microﬁltration process. Water Science and Technology: Water Supply, 6, 141–148. Ognier, S., Wisniewski, C. and Grasmick, A. (2002) Inﬂuence of macromolecule adsorption during ﬁltration of a membrane bioreactor mixed liquor suspension. Journal of Membrane Science, 209, 27–37. Ognier, S., Wisniewski, C. and Grasmick, A. (2002) Membrane fouling during constant ﬂux ﬁltration in

64

65

66

67

68

69

70

71

72

73

74

membrane bioreactors. Membrane Technology, 7, 6–10. Ognier, S., Wisniewski, C. and Grasmick, A. (2004) Membrane bioreactor fouling in sub-critical ﬁltration conditions: a local critical ﬂux concept. Journal of Membrane Science, 229, 171–177. Park, P.-K., Lee, C.-H., Choib, S.-J., Choo, K.-H., Kimd, S.-H. and Yoone, C.-H. (2002) Effect of the removal of DOMs on the performance of a coagulation-UF membrane system production. Desalination, 145, 237–245. Peuchot, M. and Ben A€ım, R. (1992) Improvement of crossﬂow microﬁltration performances with ﬂocculation. Journal of Membrane Science, 68 (3), 241–248. Rosenberger, S. and Kraume, M. (2002) Filterability of activated sludge in membrane bioreactors. Desalination, 146 (1), 373–379. Rosenberger, S., Witzig, R., Manz, W., Szewzyk, U. and Kraume, M. (2002) Performance of a bioreactor with submerged membranes for aerobic treatment of municipal waste water. Water Research, 36, 413–420. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M. and Schrotter, J.-C. (2006) Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Research, 40(4), 710–720. Seymour, R.B. and Carraher, C.E., Jr. (1988) Polymer Chemistry, Marcel Dekker Inc., New York. Sirkar, K. and Winston Ho, W.S. (1992) Membrane Handbook, Van Nostrand Reinhold, New York. Stephenson, T., Judd, S., Jefferson, B. and Brindle, K. (2000) Membrane Bioreactors for Wastewater Treatment, IWA Publishing, Great Britain. Sutherland, K. (2007) The membrane bioreactor in sewage treatment. Filtration þ Separation, 44, 18–21. Tao, G., Kekre, K., Wei, Z., Lee, T.C., Viswanath, B. and Seah, H. (2005)

References

75

76

77

78

79

80

81

82

83

Membrane bioreactors for water reclamation. Water Science and Technology, 51, 431–440. Tardieu, E., Grasmick, A., Geaugey, V. and Manem, J. (1999) Inﬂuence of hydrodynamics on fouling velocity in a recirculated MBR for wastewater treatment. Journal of Membrane Science, 156, 131–140. Ueda, T., Hata, K., Kikuoka, Y. and Seino, O. (1997) Effects of aeration on suction pressure in a submerged membrane bioreactor. Water Research, 31 (3), 489–494. Allgeier, S., Alspach, B. and Vickers, J. (2003) Membrane Filtration Guidance Manual, United States Environmental Protection Agency (EPPA), EPA 815- D- 03-008. Van den Berg, G.B. and Smolders, C.A. (1990) Flux decline in ultraﬁltration processes. Desalination, 77-1, 101–103. van der Roest, H.F., Lawrence, D.P. and van Bentem, A.G.N. (2002) Membrane Bioreactors for Municipal Wastewater Treatment, STOWA Report IWA Publishing, London. van Hoof, S.C.J.M., Duyvesteijn, C.P.T.M. and Vaal, P.P.R. (1998) Dead-end ultraﬁltration of pretreated and untreated WWTP efﬂuent for re-use in process water applications. Desalination, 118, 249–254. Wakemana, R.J. and Williams, C.J. (2002) Additional techniques to improve microﬁltration. Separation and Puriﬁcation Technology, 26, 3–18. Wen, X., Bu, Q. and Huang, X. (7–10 June 2004) Study on fouling characteristic of axial hollow ﬁbers cross-ﬂow microﬁltration under different ﬂux operations. in Proc of the IWA Specialty Conference -WEMT 2004, Seoul, Korea. Wisniewski, C., Grasmick, A. and Cruz, A.L. (2000) Critical particle size in

84

85

86

87

88

89

90

91

membrane bioreactors - case of a denitrifying bacterial suspension. Journal of Membrane Science, 178, 141–150. Wu, D.X., Howell, J.A. and Field, R.W. (1999) Critical ﬂux measurement for model colloids. Journal of Membrane Science, 152, 89–98. Xing, C.-H., Wena, X.-H. and Tardieub, E. (2001) Microﬁltration-membrane-coupled bioreactor for urban wastewater reclamation. Desalination, 141, 63–73. Xing, C.H., Wen, X.H., Qian, Y., Sun, D., Klose, P.S. and Zhang, X.Q. (2002) Fouling and cleaning of microﬁltration membrane in municipal wastewater reclamation. Water Science and Technology, 47, 263–270. Xing, C.H., Wen, X.H., Qian, Y., Wu, W.Z. and Klose, P.S. (2003) Fouling and cleaning in an ultraﬁltration membrane bioreactor for municipal wastewater treatment. Separation Science and Technology, 38, 1773–1789. Yamamoto, K., Hiasa, M., Mahmood, T. and Matsuo, T. (1989) Direct solid-liquid separation using hollow ﬁber membrane in an activated sludge aeration tank. Water Science and Technology, 21 (4/5), 43–54. Yang, W., Cicek, N. and Ilg, J. (2006) Stateof-the-art of membrane bioreactors: Worldwide research and commercial applications in North America. Journal of Membrane Science, 270, 201–211. Zeman, L.J. and Zydney, A.L. (1996) Microﬁltration and Ultraﬁltration. Principles and Applications, Marcel Dekker Inc., New York. Åhl, R.M., Leiknes, T. and Ødegaard, H. (2006) Tracking particle size distributions in a moving bed bioﬁlm membrane reactor for treatment of municipal wastewater. Water Science and Technology, 53 (7), 33–42.

j395

j397

17 Biochemical Membrane Reactors in Industrial Processes Lidietta Giorno, Rosalinda Mazzei, and Enrico Drioli

17.1 Introduction

Biochemical membrane reactors are systems able to optimally integrate and intensify chemical transformations and transport phenomena in a single unit. The transformation is promoted by a catalyst of biological origin (commonly named biocatalyst) while the transport is governed by a membrane operation (i.e., by a driving force acting through a micro-nanostructured porous or dense membrane). Transport can be appropriately tuned so as to control the reagent supply to the catalyst and/or product removal from the reaction site. The fundamentals of biochemical membrane reactors are reported in a previous chapter. Here, some highlights are just recalled for clarity in the subsequent discussion. The applications presented refer to both main reactor conﬁgurations, that is, the conﬁguration in which the membrane does not contribute to the reaction but only controls mass transport and the conﬁguration in which the reaction also occurs at the membrane level. The present work will mainly focus on biochemical membrane reactors operate at the production scale and give an overview of systems of potential interest studied at the laboratory level. Despite the various ﬁelds of application (Figure 17.1), in this work industrial sectors such as pharmaceutical, food and biotechnology will be considered. Wastewater treatment and biomedical applications are discussed in other chapters. The catalytic action of biocatalysts (enzymes, abzymes, antibodies, cells) is extremely efﬁcient and selective compared to conventional chemical catalysts. They demonstrate higher reaction rates, milder reaction conditions and greater stereospeciﬁcity. Most of these properties come from the high molecular ﬂexibility biocatalysts exhibit. On the other hand, this is also the origin of their major limit that holds back their application at the large scale, that is, the molecular stability, and then the catalyst lifetime.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 17 Biochemical Membrane Reactors in Industrial Processes

398

Figure 17.1 Field of biochemical membrane reactors application.

The use of biocatalysts in combination with membrane operations permits drawbacks to be overcome enabling biotransformation to be integrated into continuous production lines. These systems, being able to work at time-invariant conditions at steady state, permit a better control of reaction conditions with an increase of lifetime, productivity and economic viability of the process. In addition, the separation, puriﬁcation, and concentration of the obtained product can occur in a single integrated unit operation. Thanks to the biocatalyst and membrane selectivity the mass intensity can be very high, with no by-products formation, while producing high added value coproducts.

17.2 Applications at Industrial Level

Despite their great advantages, the application of biochemical membrane reactors at the industrial scale in pharmaceutical, food, and biotechnology is still limited. Major reasons for this include the nonadequate research efforts devoted to the ﬁeld, lack of predictive and holistic approach. A clear example of this situation is constituted by the commercial success submerged membrane bioreactors met in wastewater treatments. In this ﬁeld, the technology was pushed by research efforts promoted to face lack of clean water and to meet regulations about wastewater discharge in the environment. More stringent regulations about ecocompatibility of industrial processes will necessarily promote technological advances also in other industrial sectors. For example, considering the mass of wastes compared to the mass of product, it appears that pharmaceutical industry used less-advanced technology than oil reﬁneries (Figure 17.2(a)). Due to the orders of magnitude difference between the two sectors in terms of tons of productions (Figure 17.2(b)), the impact of the pharmaceutical industry is of course much lower, but it is evident that in this ﬁeld there is a much higher potential for knowledge-based technologies. Table 17.1 summarizes the most common examples of biochemical membrane reactors patented and whose robustness has been proved at the industrial production

17.2 Applications at Industrial Level

Figure 17.2 Wastes produced related to mass of product in industries [1].

scale. Table 17.1 illustrates the type of application, the biocatalyst used and the way it is used in the membrane reactor. When the enzyme is used as free, the membrane serves to separate the reaction product, whilst when it is immobilized the membrane hosts both reaction and separation. 17.2.1 Pharmaceutical Applications

The use of membrane bioreactors in the pharmaceutical ﬁeld has been documented for the production of amino acids, antibiotics, anti-inﬂammatories, anticancer drugs, vitamins, and optically pure enantiomers [10, 29–31]. Examples at the industrial scale of membrane bioreactors in the pharmaceutical ﬁeld include the production of amino acids with simultaneous regeneration of NADH, which has been commercialized by Degussa Company in Germany [10]. In Japan, the Kao Corporation investigated the so-called sandwich reactor for hydrolyzing triglycerides [32], Nitto Electric Industries immobilized cyclomaltodextrin glucanotransferasi on a hollow-ﬁber membrane and investigated production of cyclodestrins from starch [32]. Membrane bioreactors have been reported for the production of diltiazem chiral intermediate with a multiphase/extractive enzyme membrane reactor [15, 16]. The reaction was carried out in a two-separate phase reactor. Here, the membrane had the double role of conﬁning the enzyme and keeping the two phases in contact while maintaining them in two different compartments. This is the case of the multiphase/ extractive membrane reactor developed on a productive scale for the production of a chiral intermediate of diltiazem ((2R,3S)-methylmethoxyphenylglycidate), a drug used in the treatment of hypertension and angina [15]. The principle is illustrated in

j399

j 17 Biochemical Membrane Reactors in Industrial Processes

400

Table 17.1 Biochemical membrane reactors in industrial processes.

Biocatalyst

Status

Application

Lactase

Immobilized

Glucose isomerase

Immobilized

Penicillin acylase Acylase E. Coli Pseudomonas dacunahe Aminoacilase, and dehidrogenase Brevibacterium ammoniagenes Pectic enzymes

Free Immobilized Immobilized Immobilized

Hydrolysis of beta-D-galactosidic linkage of lactose milk (Industrial scale) [2] Conversion of D-glucose to D-fructose (Industrial scale) [3] Production of antibiotics (Industrial scale) J. [4] Production of L-aminoacids (Industrial scale) [5] Production of L-aspartic acid (Industrial scale) [6] Production of L-alanine (Industrial scale) [7]

Free and immobilized Immobilized

Thermolysin Lipase of 360

Free or immobilized Immobilized Immobilized

Trypsin

Free

Protease Acetyl transferase from Taxus Lipase Cells

Immobilized Immobilized Free —

Pancreatic cells

Immobilized

Cells Stem cells Viruses, virus particles, antibodies and proteins Cells

Free — Free

Production of L-aminoacids [8–10] Production of L-malic acid (Industrial scale) [11] Hydrolysis of pectins to improve processability (industrial scale) [12, 13] Production of aspartame (Industrial scale) [14] Production of diltiazem chiral intermediate (industrial scale) [15, 16] Production of casein bioactive peptide (patented) [17, 18]. Hydrolysis of caroteno-proteins (patented) [19] Production of baccatin III (patented) [20] Production of fatty acid (patented) [21] Linear or membrane-like biodevices and a bioreactor in which adhesive cells are anchored at high density (patented) [22] Artiﬁcial organs and implantable bioreactors (patented) [23, 24] Continuous cell culture (patented) [25] Cell-expansion apparatus (patented) [26] Production of a concentrated solution from biological substances (patented) [27] Delivery of drugs or genes to individual cells (patented) [28]

Figure 17.3, the reactant was fed into the solvent, while the product was extracted in water. The lipase was immobilized by entrapment method on asymmetric PAN hollow-ﬁber membranes. The process was run for several years with modules for the production plant of 60 m2 of active membrane area. A further improvement of the multiphase reactor concept using lipase for enantioselective transformation has been recently reported, that is, an emulsion enzyme membrane reactor. Here, the organic/water interface within the pores at the enzyme level is achieved by stable oil-in-water emulsion, prepared by membrane emulsiﬁcation. In this way, each pore forms a microreactor containing immobilized

17.2 Applications at Industrial Level

Figure 17.3 Schematic representation of multiphase membrane reactor.

enzyme (Figure 17.4) [33]. In the membrane pores, the enzyme is able to work in the same conditions as in the stirred-tank reactor, but with no shear stress due to stirring. This conﬁguration improved the selectivity and productivity of the biocatalytic system as well as its catalytic stability, conﬁrming that the observed inversion relationship between activity and stability of immobilized enzyme is not a general rule. Other biochemical membrane reactors applications include the synthesis of lovastatin with immobilized Candida rugosa lipase on a nylon support [34]; the synthesis of isomalto oligosaccharides and oligodextrans in a recycle membrane bioreactor by the combined use of dextransucrase and dextranase [35], the production of a derivative of kyotorphin (analgesic) in solvent media using a-chymotrypsin as catalyst and a- alumina mesoporous tubular support [36], and biodegradation of high-strength phenol solutions by Pseudomonas putida using microporous hollow ﬁbers [37]. A particular application of membrane bioreactors, patented in 2005 [20], concerns the production of an antitumor substance (paclitaxel). Since a full synthesis of paclitaxel is not possible due to its low yield, a semisynthesis of 10-deacetyl-baccatin

Figure 17.4 Representation of emulsion enzyme membrane.

j401

j 17 Biochemical Membrane Reactors in Industrial Processes

402

III (10-DAB) was carried out from which baccatin III was produced in an enzyme reactor. The enzyme reactor comprised a hollow-ﬁber polymeric ultraﬁltration membrane, with immobilized acetyl transferase from Taxus species. The process enabled the production of baccatin III without requiring complicated puriﬁcation steps of the acetyl transferase. The puriﬁcation of the baccatin III is also made distinctly easier [20]. Membrane bioreactors can be easily integrated with other systems, for example, with delivery of drugs or genes to individual cells achieved on the nanoscale using electroporation techniques. In one method developed in a recent patent, a ﬂowthrough bioreactor having an inlet and an outlet connected by a ﬂow chamber and a nanoporous membrane positioned in the ﬂow chamber was used [28]. Recent studies in the pharmaceutical ﬁeld using MBR technology are related to optical resolution of racemic mixtures or esters synthesis. The kinetic resolution of (R,S)-naproxen methyl esters to produce (S)-naproxen in emulsion enzyme membrane reactors (E-EMRs) where emulsion is produced by crossﬂow membrane emulsiﬁcation [38, 39], and of racemic ibuprofen ester [40] were developed. The esters synthesis, like for example butyl laurate, by a covalent attachment of Candida antarctica lipase B (CALB) onto a ceramic support previously coated by polymers was recently described [41]. An enzymatic membrane reactor based on the immobilization of lipase on a ceramic support was used to perform interesteriﬁcation between castor oil triglycerides and methyl oleate, reducing the viscosity of the substrate by injecting supercritical CO2 [42]. The production of aromatic compounds by a membrane bioreactor is widely studied and some examples are also patented. Aromatic compounds are important substances in pharmaceutical, food, and cosmetic industries due to their natural properties and because they are strong antioxidant molecules with a strong freeradical scavenging activity. The hydrolysis of caroteno-proteins for the production of astaxanthin using protease was developed in an enzymatic membrane bioreactor [19], in which the concentration of the protein fraction by ultraﬁltration and the separation of the pigments in the permeate were simultaneously carried out. Terpene esters belong to a large family of aromatic compounds, which are important for ﬂavoring and are widely used in the pharmaceutical and food industries. An important terpene ester is a-pinene oxide, its biotransformation to isonovalal using resting cells of Pseudomonas ﬂuorescens NCIMB 11671 was evaluated in a membrane bioreactor [43]. Production of geranyl acetate, one of the best-known aromatic compounds, was studied using lipase CAL-B immobilized on polymer membranes by sorption and chemical binding [44]. 17.2.2 Food Applications

The main applications of biocatalytic membrane reactors in the food sector include: reduction of the viscosity of juices by hydrolyzing pectins, reduction of the lactose content in milk and whey by its conversion into digestible sugar, treatment of musts

17.2 Applications at Industrial Level

and wine by the conversion of polyphenolic compounds and antocyanes and the removal of peroxides from diary products. The interactions between pectins and sugars (rhamnose, arabinose, and galatose) are principally responsible for the high turbidity and viscosity of fruit juice. Pectinases immobilized in membranes are used to reduce the viscosity of fruit juice [12, 13]. One of the ﬁrst cases of the application of membrane bioreactors in food processes was the production of milk with low lactose content. b-galactosidase was entrapped into cellulose acetate ﬁbers to carry out the hydrolysis of milk and whey lactose [2] recently the system was improved by the use of microﬁltration and by UV irradiation of the enzyme solution to avoid growth of micro-organisms [45]. The use of membrane reactors as continuous systems for the hydrolysis of lactose (present in whole milk or cheese whey) is an effective technique running at a large scale. Intolerance to milk is not only caused by lactose, but also by high molecular weight proteins. In fact, some children and old people have difﬁculty in hydrolyzing proteins with molecular weight higher than 5 kDa. In other words, they cannot digest such proteins, which induce stomach ache and can also lead to allergy. The hydrolysis of high molecular weight proteins into polypeptides lower than 5 kDa in biocatalytic membrane reactors is a new approach to produce low allergenic fresh milk with improved properties compared to the reconstituted powder milk currently used. The biocatalytic membrane reactor can be designed so that the hydrolyzed fragments equal to or lower than 5 kDa can be removed through a membrane of appropriate cutoff, while retaining the nonhydrolyzed proteins. In order to achieve high efﬁciency, the hydrolytic step should be part of an integrated system where up- and downstream of milk is properly considered. Biocatalytic membrane reactors can also be used to valorize coproducts of cheese-making processes. In fact, it is possible to increase the cost effectiveness of cheese-making processes and reduce waste simply by recovering and reusing compounds present in waste streams. The whey proteins (such as a-lactalbumin), which have excellent functional properties, can be recovered by ultraﬁltration and hydrolyzed to produce many useful pharmaceutical intermediates. In addition, permeates from the ultraﬁltered milk and whey contain lactose, which can be recovered and used in the production of glucose and galactose syrup. Other important applications in the food industry running at a large scale are the production of L-aspartic acid with Escherichia coli entrapped in polyacrilamides [6], the immobilization of thermolysin for the production of aspartame [14], The production of L-alanine by Tanabe Seiyaku [7], the production of fructose concencentrated syrup [3], the production of L-malic acid by the use of Brevibacterium ammoniagenens immobilized in polyacrilamide by entrapment immobilization methods [11] and L-aminoacids production by immobilized aminoacylase [5]. Biocatalytic membrane reactors are also used for the treatment of musts and wines by the conversion of polyphenolic compounds and anthocyanes. Laccase is used to oxidize polyphenols in solution and anthocianase is used immobilized on synthetic and natural polymers to hydrolyze anthocyanes. During the maturation process, a secondary fermentation occurs that converts malic acid into lactic acid. Control of this reaction will enable the production of

j403

j 17 Biochemical Membrane Reactors in Industrial Processes

404

a product with good organoleptic properties. In comparison with batch fermentation, membrane bioreactors achieve remarkably elevated cell concentrations and productivity. In this type of bioreactor, a membrane has been installed to prevent washout of yeast when the fermented stream is withdrawn from the reactor. This apparatus makes it possible to keep the cell concentration in the reactor high while reducing product inhibition by replacing product-containing broth with fresh medium. Recent patented works were also reported related to the design of membrane bioreactors. A membrane biological reactor of a ﬂat x-plate system comprising a thermostat, and a free system comprising a hydrophilic membrane with cutoff value of 30 100 kDa at 15–65 C using lipase enzyme was developed and patented for the production of fatty acids by enzymatic hydrolysis of vegetable or animal oils or fats [21]. Membrane bioreactors were also used to improve existing systems for different applications, like cells nutrition and growth, and as delivery systems for genes and drugs [22–25, 28], to produce concentrated solution consisting of biological substances such as, for example, viruses, virus particles, antibodies, and proteins [27]. The design of food products that confer a health beneﬁt is a relatively new trend, and recognizes the growing acceptance of the role of diet in disease prevention, treatment, and well-being. This change in attitude for product design and development has forced organizations and industries involved in formulating foods for health beneﬁt into new areas of knowledge. Recent works in the literature were devoted to improve the production of food similar to an existing one that is less dangerous in a compromised diet for the production of functional food or the production of nutraceuticals. Palatinose (isomaltulose, 6-O-a-D-glucopyranosyl-D-fructose) a natural substance with a sweetening power of about 45% less than that of sucrose, and xilitol, where insulin is not involved in its methabolism, can be used as sugar substitutes in diabetic subjects. Recently a complete conversion of concentrated sucrose solutions into palatinose immobilizing in a hollow-ﬁber membrane reactor Serratia plymuthica cells was produced [46]. Candida tropicalis, an osmophilic strain isolated from honeycomb, was used to produce xilitol recycled in a submerged membrane bioreactor with suction pressure and air sparging, obtaining the highest productivity of xilitol 12.0 g l1 h1 [47]. Octenylscuccinate derivatives of starch are attracting growing attention of food technologists as potential additives as emulsifying agents. The results obtained using the technology of membrane bioreactors indicated that the hydrolysis of sodium octenylsuccinate starch derivatives leads to products that reveal the surface activity, irrespective of the type as well as the amount of enzyme used in the hydrolysis process. The application of the UF membrane bioreactor to enzymatic hydrolysis could be the way of intensiﬁcation of the production process [48]. A very interesting ﬁeld in membrane bioreactors is the production of cyclodextrins or oligosaccharides. In general, they have applications in food pharmaceutical, cosmetic agricultural, and plastics industries as emulsiﬁers, antioxidant, and stabilizing agents. In the food industry cyclodextrins are employed for the preparation of cholesterol-free products. The use of enzymatic membrane reactors to produce

17.2 Applications at Industrial Level

cyclodextrins has been reported starting from different sources including soluble potato and corn starch. A recent study [49] reported also their production using enzyme membrane reactors starting from tapioca starch. The production of oligosaccharides to be used as functional food was also obtained by the immobilization of dextranase on polymeric matrix [50]. The production of substances that preserve the food from contamination or from oxidation is another important ﬁeld of membrane bioreactor. For example, the production of high amounts of propionic acid, commonly used as antifungal substance, was carried out by a continuous stirred-tank reactor associated with ultraﬁltration cell recycle and a nanoﬁltration membrane [51] or the production of gluconic acid by the use of glucose oxidase in a bioreactor using PES membranes [52]. Lactic acid is widely used as an acidulant, ﬂavor additive, and preservative in the food, pharmaceutical, leather, and textile industries. As an intermediate product in mammalian metabolism, L( þ ) lactic acid is more important in the food industry than the D() isomer. The performance of an improved fermentation system, that is, a membrane cell-recycle bioreactors MCRB was studied [53, 54], the maximum productivity of 31.5 g/L h was recorded, 10 times greater than the counterpart of the batch-fed fermentation [54]. 17.2.3 Immobilization of Biocatalysts on Membranes

The choice of reactor conﬁguration depends on the properties of the reaction system. For example, bioconversions for which the homogeneous catalyst distribution is particularly important are optimally performed in a reactor with the biocatalyst compartmentalized by the membrane in the reaction vessel. The membrane is used to retain large components, such as the enzyme and the substrate while allowing small molecules (e.g., the reaction product) to pass through. For more labile molecules, immobilization may increase the thermal, pH and storage stability of biocatalysts. Biocatalysts can be entrapped within the membrane, geliﬁed on the membrane surface or bounded to the membrane surface or inner polymeric matrix [55]. The entrapment method of immobilization is based on the localization of an enzyme within a polymer membrane matrix. It is done in such a way as to retain biocatalyst, while allowing penetration of substrate. Asymmetric hollow ﬁbers provide an interesting support for enzyme immobilization, in this case the membrane structure allows the retention of the enzyme into the sponge layer of the ﬁbers by crossﬂow ﬁltration. The amount of biocatalyst loaded, its distribution and activity through the support and its lifetime are very important parameters to properly orientate the development of such systems. The speciﬁc effect that the support has upon the enzyme, however, greatly depend upon both the support and the enzyme involved in the immobilization as well as the method of immobilization used. The amount of the immobilized biocatalyst is an important parameter and strongly affects the reactor performance, enzymes in fact, are not able to work at high concentration.

j405

j 17 Biochemical Membrane Reactors in Industrial Processes

406

The amount of immobilized protein can be determined by the mass balance between initial and ﬁnal solutions [56]. A combined qualitative method merged from the classical in-situ detection of enzyme activity and western blot analysis can be applied to determine the enzyme spatial distribution through the membrane thickness and along the membrane module and its activity after the immobilization [57–59]. The geliﬁcation of the biocatalyst on the membrane is based on one of the main drawbacks of membrane processes: fouling. Disadvantages of this systems are the reduction of the catalytic efﬁciency, due to mass transport limitations and the possibility of preferential pathways in the enzyme gel layer [60]. The binding of the biocatalyst to the membrane can be divided into three principal groups: ionic binding, crosslinking, and covalent linking. In the literature there are various routes to carry enzyme immobilization creating a bound on supports, the principal strategies are based on chemical grafting or molecular recognition on porous supports. The sites involved in this chemistry are generally carboxylic acid, hydroxyls, amino or quaternary ammonium groups, which are created on the surface of porous material by various means, like direct chemical surface treatment or plasma or UV activation. The reactive sites thus created allow the attachment of the enzyme by use of coupling reagents such as tosyl chloride, dicyclohexylcarbodiimide and glutaraldehyde. Approaches aiming at creating biocompatible environments consist in modifying the surface of polymeric membranes by attaching functional groups like sugars, polypeptides and then to adsorb the enzymes. Another way considered as of biomimetic inspiration and that was shown to be efﬁcient for enzyme attachment, it consists in using the very strong and speciﬁc interaction of the small protein avidin for the biotin [61, 62]. The tetrameric structure of avidin permits itself to interact with four different molecules of biotin at the same time. Various proteins and enzyme could be easily biotinylated, and this mode of enzyme grafting has already been used for electrodes production as well as for membranes made up of conducting ﬁbers. Although immobilization of enzymes generally enhance their stability, one major disadvantage of random immobilization of enzymes onto polymeric microﬁltrationtype membranes is that the activity of the immobilized enzymes is often signiﬁcantly decreased because the active site may be blocked from substrate accessibility, multiple-point binding may occur, or the enzyme may be denatured [31]. Different approaches are developed in order to accommodate site-speciﬁc immobilization of enzymes with different structural characteristics, as gene fusion to incorporate a peptidic afﬁnity tag at the N- or C- terminus of the enzyme; post-translational modiﬁcation to incorporate a single biotin moiety on enzymes; and site-directed mutagenesis to introduce unique cysteins to enzymes [63]. The selection of the membrane to be used in enzymatic membrane reactors should take into account the size of the (bio)catalyst, substrates, and products as well as the chemical species of the species in solution and of the membrane itself. An important parameter to be used in this selection is the solute-rejection coefﬁcient, which should

References

be zero for the product to facilitate permeation, and should be one for the enzyme to insure a complete retention of the catalyst inside the reaction system. The selectivity is normally associated with a discrimination based on size exclusion, but when a steric exclusion process may be present for molecules with size close to the pore size.

17.3 Conclusion

Currently, the major industrial application of biochemical membrane reactors is in water treatment. In the biotechnology ﬁeld the development of such biohybrid catalytic systems is still at an emerging stage. The main technological difﬁculties in using membrane bioreactors for production on an industrial level are related with rate-limiting aspects and reproducibility on the large scale, together with the life-time of the enzyme, the availability of pure catalysts at an acceptable cost, and the necessity for biocatalysts to operate at low substrate concentration and without microbial contamination. Many studies are oriented to the investigation of operating conditions and optimization of the various properties of membrane bioreactors. However, efforts towards the development of a predictive knowledge-based approach, able to overcome the trial and error one, is necessary to signiﬁcantly advance the ﬁeld. The development of membranes speciﬁcally designed to answer bioreactor needs is needed as well. The converging of technologies such as genetic engineering, bioprocess design, molecular modeling and biochemical engineering will promote innovative solutions to face the need for precise, selective, clean, safe, low energy consumption and ecofriendly processes, such as biochemical membrane bioreactors.

References 1 Sheldon, R.A. (1997) Journal of Chemical Technology and Biotechnology, 68, 381. 2 Pastore, M. and Morisi, F. (1976) Methods in Enzymology, 44, 822–830. 3 Carasik, W. and Carrol, J.O. (1983) Food Technology, 37, 85–91. 4 Bryjak, J., Bryjak, M. and Noworyta, A. (1996) Enzyme and Microbial Technology, 19, 196–201. 5 Sato, T. and Tosa, T. (1993) Optical resolution of aminoacids by aminoacylase in Industrial Application of Immobilized Biocatalyst (eds T. Tanaka, T. Tosa and T. Kobayashi) Marcel Dekker, New York, pp. 3–14.

6 Chibata, I. et al. (1974) Applied Microbiology, 27, 878–885. 7 Takamatsu, S. (1982) European Journal of Applied Microbiology and Biotechnology, 15, 147–149. 8 Wandrey, C. and Flaschel, E. (1979) Advances in Biochemical Engineering, vol. 12, Springer Verlag, Berlin, pp. 147. 9 Wandrey, C., Wichmann, R., Leuchtenberger, W. and Kula, M.R. (1981) US Patent No. 4,304,858, Degussa AG/GBF. 10 Wandrey, C. (2004) Biochemical reaction engineering for redox reactions. The Chemical Record, 4, 254–265.

j407

j 17 Biochemical Membrane Reactors in Industrial Processes

408

11 Takata, I., Tosa, T. and Kobayashi, T. (eds.) (1993) Industrial Application of Immobilized Biocatalysts, Marcel Dekker, New York, pp. 53–55. 12 Alkorta, I., Garbisu, C., Llama, M.J. and Serra, J.L. (1998) Process Biochemistry, 33, 21–28. 13 Giorno, L. et al. (1998) Fruit Processing Separation Science and Technology, 33, 739–756. 14 Oyama, K. (1974) Journal of Chemical Society, 11, 356–360. 15 Lopez, J.L. and Matson, S. (1997) Journal of Membrane Science, 125, 189–211. 16 Matson, S.L. (1989) Patent No. US4800162. 17 Qi, H.Z. (2004) Patent No. CN1546682. 18 Trusek-Holownia, A. (2008) Biochemical Engineering Journal, 39, 221–229. 19 Guerrero, L.M.I. (2004) Patent No MXPA02012838. 20 Frense, D. (2005) Patent No WO2005066353. 21 Belaﬁne Bako, K. (2006) Patent No HU0401348. 22 Satoshi, K. and Tetsuya, M. (2006) Patent No CA2511457. 23 Shalev, A. (2006) Patent No WO2006080009. 24 Shalev, A. (2008) Patent No US2008112995. 25 Zhang, Z., Boccazzi, P., Choi, H., Jensen, K. F. and Sinskey, A. (2006) Patent No. US2006199260. 26 Antwiler Glen, D. (2008) WO2008109200. 27 Czermak, P. and Nehring, D. (2006) Patent No WO2006005305. 28 Lee, L.J., Wang, S., Xie, Y., Zeng, C., Koh, C.G. and Fei, Z. (2007) Patent No WO2007053802. 29 Giorno, L. and Drioli, E. (2000) Trends in Biotechnology, 18, 339–348. 30 Giorno, L., Li, N. and Drioli, E. (2003) Biotechnology and Bioengineering, 84, 677. 31 Charcosset, C. (2006) Biotechnology Advances, 24, 482–492. 32 Nam, C.H. and Furusaki, S. (1991) Membrane bioreactors: Present and

33

34 35

36

37

38

39 40

41

42

43

44 45 46 47

48

prospects. Advances in Biochemical Engineering/Biotechnology, 44, 27–64. Giorno, L., DAmore, E., Mazzei, R., Piacentini, E., Zhang, J., Drioli, E., Cassano, R. and Picci, N. (2007) Journal of Membrane Science, 295, 95–101. Yang, W., Cicek, N. and Ilg, J. (1997) Journal of Membrane Science, 270, 201–211. Goulas, A.K., Cooper, J.M., Grandison, A.S. and Rastall, R.A. (2004) Biotechnology and Bioengineering, 88, 778–787. Belleville, M.P., Lonzano, P., Iborra, J.L. and Rios, G.M. (2001) Separation and Puriﬁcation Technology, 25, 229–233. Chung, T.P., Wu, P.C. and Juang, R.S., (2005) Journal of Membrane Science, 317, 55–63. Giorno, L., Piacentini, E., Mazzei, R. and Drioli, E. (2008) Journal of Membrane Science, 317, 19–25. Li, N. and Sakaki, K. (2008) Journal of Membrane Science, 314, 183–192. Long, W.S., Azlina Harun, K. and Subhash, I. (2005) Chemical Engineering Science, 60, 4957–4970. Gumı, T., Fernandez-Delgado Albacete, J., Paolucci-Jeanjean, D., Belleville, M. and Rios, G.M. (2008) Journal of Membrane Science, 311, 147–152. Pomier, E., Delebecque, N., PaolucciJeanjean, D., Pina, M., Sarrade, S. and Rios, G.M. (2007) Journal of Supercritical Fluids, 41, 380–385. Boontawan, A. and Stuckey, D.C. (2006) Applied Microbiology and Biotechnology, 69, 643–649. Trusek-Holownia, A. and Noworyta, A. (2007) Journal of Biotechnology, 130, 47–56. Novalin, S., Neuhaus, W. and Kulbe, K. (2005) Journal of Biotechology, 119, 212–218. Krastanov, A., Blazheva, D. and Stanchev, V. (2007) Process Biochemistry, 42, 1655–1659. Kwona, S., Park, S. and Oh, D. (2006) Journal of Bioscience and Bioengineering, 101, 13–18. Prochaskaa, K., Kedzioraa, P., Le Thanhb, J. and Lewandowicz, G. (2007) Food Hydrocolloids, 21, 654–659.

References 49 Sakinaha, A.M.M., Ismaila, A.F., Md Illiasa, R. and Hassand, O. (2007) Desalination, 207, 227–242. 50 Torras, C., Nabarlatz, D., Vallot, G., Montane, D. and Garcia-Valls, R. (2008) Chemical Engineering Journal, 144, 259–266. 51 Boyaval, P. and Corre, C. (1995) Le Lait, 75, 453–461. 52 Liu, J. and Cui, Z. (2007) Journal of Membrane Science, 302, 180–187. 53 Giorno, L., Chojnacka, K., Donato, L. and Drioli, E. (2002) Industrial & Engineering Chemistry Research, 41, 433. L–440.L. 54 Xu, G., Chu, J., Wang, Y., Zhuang, Y., Zhang, S. and Peng, H. (2006) Process Biochemistry, 41, 2458–2463. 55 Strathmann, H., Giorno, L. and Drioli, E. (2006) An introduction to membrane science and technology, CNR, Rome. 56 Giorno, L. et al. (1995) Journal of Chemical Technology and Biotechnology (Oxford, Oxfordshire: 1986), 64, 345–352.

57 Mazzuca, S., Giorno, L., Spadafora, A., Mazzei, R. and Drioli, E. (2006) Journal of Membrane Science, 285, 152–158. 58 Crespo, J.P.S.G., Trotin, M., Hough, D. and Howell, J.A. (1999) Journal of Membrane Science, 155, 209. 59 Liu, Z.M., Dubremez, J., Richard, V., Yang, Q., Xu, Z.K. and Seta, P. (2005) Journal of Membrane Science, 267, 2. 60 Drioli, E. and Iorio, G. (1989) Enzyme membrane reactor and membrane fermentators, Handbook of industrial membrane technology (ed. C.P. Mark), Noyes Publications, Park Ridge, New Jersey, USA, pp. 401–481. 61 Amounas, M., Innocent, C., Cosnier, S. and Seta, P. (2000) Journal of Membrane Science, 176, 169–176. 62 Amounas, M., Magne, V., Innocent, C., Dejean, E. and Seta, P. (2002) Enzyme and Microbial Technology, 31, 171–178. 63 Butterﬁeld, D.A., Bhattacharyya, D., Daunert, S. and Bachas, L. (2001) Journal of Membrane Science, 181, 29–37.

j409

j411

18 Biomedical Membrane Extracorporeal Devices Michel Y. Jaffrin and Cecile Legallais

18.1 General Introduction 18.1.1 Use of Membranes in the Medical Field

Medical applications of membranes are a fast-growing ﬁeld that represents the largest consumption of membrane area per year. The ﬁrst and most important application in terms of cost is the treatment of end-stage renal disease (ESRD) by hemodialysis that consumes about 108 m2 of membranes per year, followed by blood oxygenators used during cardiac surgery and in the case of respiratory failure with more than 5 106 m2 per year. Plasma separation and fractionation used for plasma collection from donors and in the treatment of autoimmune diseases and cholesterol removal account for another 2.5 106 m2 per year. New membrane applications in artiﬁcial and bioartiﬁcial organs, for instance for liver and pancreatic function support, are presently emerging. This chapter will focus on three types of membrane extracorporeal devices, hemodialyzers, plasma ﬁlters for fractionating blood components, and artiﬁcial liver systems. These applications share the same physical principles of mass transfer by diffusion and convection across a microﬁltration or ultraﬁltration membrane (Figure 18.1). A considerable amount of research and development has been undertaken by membrane and modules manufacturers for producing more biocompatible and permeable membranes, while improving modules performance by optimizing their internal ﬂuid mechanics and their geometry. 18.1.2 Historical Perspective

Hemodialysis was ﬁrst used to treat wounded British pilots from 1943 to 1945 by W. Kolff at the Groningen hospital in Holland. The patients blood was circulated

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 18 Biomedical Membrane Extracorporeal Devices

412

Figure 18.1 Mass transfer or removal methods available in the field of artificial organs.

inside a ﬂat cellulose tubing normally used for wrapping sausages, wound around a wooden drum with an horizontal axis, which was rotated above a ﬂat tank containing an ionic solution. This principle of a spiral wound membrane was retained in the ﬁrst generation of coil-type disposable dialyzers produced from 1955 to 1970. The real development of hemodialysis was permitted by the arteriovenous shunt proposed by Scribner (Seattle, USA) in 1960 that permitted an easier blood access with an increased blood ﬂow into the dialyzer. In 1966, this external shunt was replaced by the arteriovenous ﬁstula of Ciminio (Brescia, Italy) implanted under the skin that was safer and more comfortable for the patient. Press-type disposable dialyzers replaced coil types in 1969 and the ﬁrst hollow-ﬁber modules, more compact and lighter than the press-type ones, were introduced in 1972. The need to transfuse blood components such as plasma, platelets, factor VIII, in addition to red blood cells (RBC) has generated the development of plasmapheresis (plasma separation from whole blood) and more generally that of apheresis (fractionation of blood components). Plasma collection from donors by centrifugation of blood bags began only in 1944. This technique was extended to therapeutic plasma puriﬁcation in 1950, but RBCs were fragilized by the centrifugation and the plasma was not completely platelet-free. Plasma separation by membrane microﬁltration was proposed in 1978 by Salomon et al. [1] as a substitute to centrifugation and its clinical potential conﬁrmed in 1980 by Samtleben et al. [2]. This technique yields a high-quality cell-free plasma that avoids for the recipient the immunological hazards of contamination by platelets and cellular fragments and is less traumatic for red cells, if precautions are taken to avoid hemolysis during ﬁltration. In the meanwhile, other organ replacements were investigated. In the case of liver supply, both artiﬁcial and bioartiﬁcial (using hepatic cells) approaches proposed in

18.2 Hemodialyzers

the 1960s and 1970s [3, 4] did not succeed due to the livers complexity. Now, better insights into the organ physiology help in designing new and sophisticated techniques able to remove protein-bound toxins and, in the bioartiﬁcial case, to supply biotransformation, storage and synthesis functions.

18.2 Hemodialyzers 18.2.1 Introduction

Hemodialysis (HD) permits treatment of end-stage renal failure (ESRD) in patients with a residual renal capacity of less than 10% of normal one. It consists in an extracorporeal circulation of the patients blood withdrawn from a vein through a needle, which circulates in a hemodialyzer before being returned to the patient (Figure 18.2a). The hemodialyzer extracts uremic toxins from the blood by diffusion across an ultraﬁltration membrane with a cutoff of no more than 30 kDa so as to retain red blood cells and plasma proteins. These toxins are carried away by an ionic solution (the dialysate) that circulates at counter current to blood on the other side of the membrane. Treatment of ESRD by HD generally requires 3 four-hour dialysis runs per week as illustrated in Figure 18.2b and concerns now more than 1 700 000 patients worldwide. Another method of treatment, peritoneal dialysis (PD) consists in ﬁlling the peritoneal cavity with dialysate and does not necessitate an extracorporeal blood circulation as the hemodialyzer is replaced by the peritoneal membrane. Statistics of various modes of ESRD treatment are given in Table 18.1 [5]. Since hemodialyzers are used only once or reused up to 10 times, but no more than in

Figure 18.2 (a) Schematic of extracorporeal circuit in hemodialysis and hemodiafiltration. Substitution fluid is added only in the case of large UF flow rate, that is, in HDF. (b) Fresenius monitor 5008 used for HD or HDF.

j413

j 18 Biomedical Membrane Extracorporeal Devices

414

Table 18.1 Statistics of various types of ESRD treatment.

Treatment type

France

Europe

World

Hemodialysis Peritoneal dialysis HDF, HF Total ESRD

26 000 3000 4000 33 000

260 000 30 000 50 000 340 000

1300 000 160 000 440 000 1900 000

HDF: hemodiaﬁltration. HF: hemoﬁltration [5].

a few countries, hemodialysis consumes more m2 of membrane area than all other membrane applications combined. Fortunately, due to automatized production, hemodialyzers cost from D 15 to 30 a piece, which is much less per m2 than most industrial membranes, except large spiral-wound desalination modules, when purchased in large quantities. Hemodialyzers are generally of hollow-ﬁber types with an internal diameter of about 200 mm and a membrane area from 1.2 to 2 m2. Current research is focused on augmenting the efﬁciency of large toxin removal by using convective transfer in addition to diffusion and on improving membrane biocompatibility. Other goals are the development of hemodiaﬁltration (HDF) a combination of hemodialysis with large ultraﬁltration [6] by reducing its cost with online production of reinfusion ﬂuid by microﬁltration of dialysate by the generator [7]. 18.2.2 Physical Principles of Hemodialysis

As blood circulates along the dialyzer membrane, uremic toxins diffuse into the dialysate that is discarded, under the action of the concentration gradient (Figure 18.3).

Figure 18.3 Schematic of fluid and mass transfer between blood and dialysate compartments in a hemodialyzer.

18.2 Hemodialyzers

To prevent the blood from losing its ions, these ions are included in the dialysate at the same concentration as in normal plasma to stop their diffusion through the membrane. Since uremic patients produce little or no urine, their excess water is eliminated from the blood by ultraﬁltration through the membrane since blood pressure is higher than dialysate pressure. This ultraﬁltration also contributes to toxin removal by carrying toxins into the dialysate. This convective transfer is more efﬁcient than diffusion for high molecular weight toxins.

18.2.3 Dialysis Requirements

The dialyzer must carry out four tasks [8]: (a) The elimination of uremic toxins (urea, creatinin, uric acid, phosphates, etc.) mainly by diffusion in normal HD, but also by convection in HDF; (b) The elimination of excess water, generally from 1 to 4 l, by ultraﬁltration; (c) The regulation of plasmatic ion concentrations (Na þ , K þ , Ca þ þ , Mg þ þ ) by dialysate composition; (d) The regulation of pH around 7.2 to avoid acidosis by introducing a buffer, acetate or sodium bicarbonate, in the dialysate. The ion concentrations of normal plasma and of the plasma in the case of renal insufﬁciency are compared in Table 18.2 with that of dialysate. The dialysate contains fewer K þ and Mg þþ ions than normal plasma, but more HCO3. The pathologic plasma contains on average 6 times more urea and 10 times more creatinin than the normal one.

Table 18.2 Comparison of ion, urea, creatinin, glucose, and protein concentrations of normal plasma and plasma in case of renal insufficiency with those of dialysate.

Solute

Dialysate

Normal plasma

ESRD plasma

Naþ (mmol/L) Kþ (mmol/L) Mg2þ (mmol/L) Ca2þ (mmol/L) Cl (mmol/L) HCO3 (mmol/L) Urea (mmol/L) Creatinin (mmol/L) Glucose (g/L) Proteins (g/L)

140 2 0.5 1.75 106.5 35 0 0 0 0

142 4 1.5 2.5 103 27 5 0.1 1.0 70

140 5.5 1.5 2 103 21 30 1.0 1.0 70

j415

j 18 Biomedical Membrane Extracorporeal Devices

416

18.2.4 Mass Transfers in a Hemodialyzer

Let us ﬁrst consider, for simplicity, the case of diffusive transfer without ultraﬁltration. The local mass ﬂux through the membrane per unit area Js (x) of a speciﬁc toxin is given by Js ðxÞ ¼ ðCB ðxÞCD ðxÞÞ=RT

ð18:1Þ

where CB(x) and CD(x) denote, respectively, the bulk toxin concentrations in blood and dialysate at distance x from the dialyzer inlet and RT is the sum of diffusive resistances in the blood, membrane, and dialysate [9], RT ¼ RB þ Rm þ RD

ð18:2Þ

18.2.4.1 Characterization of Hemodialyzers Performance The capacity for toxin removal of an hemodialyzer is expressed by the dialysance or the clearance which have the same unit as a ﬂow rate. The dialysance D is deﬁned, for each toxin as

D¼

toxin mass removed per unit time Q Bi CBi Q Bo CDo ¼ CBi CD CBi CD

ð18:3Þ

where QB denotes the blood ﬂow rate, the subscripts i and o denote, respectively, the inlet and outlet of the dialyzer. The clearance K is a particular case of the dialysance and is used when the inlet dialysate does not contain any toxin (CDi ¼ 0) which is the case of normal dialysate generators in which the dialysate circulates in open circuit. Thus, from Equation 18.3 K¼

Q Bi CBi Q Bo CBo CBi

ð18:4Þ

Equations 18.3 and 18.4 are valid when an ultraﬁltration ﬂow rate Q F is present and inlet and outlet blood ﬂow rates are related by Q Bi ¼ Q Bo þ Q F

ð18:5aÞ

Similarly, for the dialysate ﬂow rate Q D, we have Q Di ¼ Q Do Q F

ð18:5bÞ

If the ultraﬁltration is small, as in normal hemodialysis, Q F Q Bi, Q Bi Q Bo ¼ Q B and Equation 18.4 becomes K Q B ð1CBo =CBi Þ

ð18:6Þ

The clearance can be calculated in the absence of ultraﬁltration (diffusive clearance KD) by writing mass balances equations in the blood and dialysate phases and using Equation 18.6. The result is [9] KD ¼ Q B

ea 1 B =Q D

ea Q

ð18:7Þ

18.2 Hemodialyzers

Figure 18.4 Variations of clearances of various solutes with blood flow.

with S 1 1 a¼ RT Q B Q D

ð18:8Þ

where S is the membrane area. This diffusive clearance increases when QB and QD increase and when the molecular weight of the toxin decreases as RT decreases. Due to the arteriovenous shunt, the blood ﬂow rate withdrawn from the patients vein is between 200 and 400 ml/min, while the dialysate ﬂow rate is generally set by the generator at 500 ml/min. A typical variation of the clearance of various toxins with blood ﬂow rate is shown in Figure 18.4. Equations 18.7 and 18.8 indicate that the diffusive clearance is independent of toxin concentration and will not vary during dialysis, if ﬂow rates are constant. If CDi 6¼ 0, when dialysate recirculates in a closed circuit or when considering the transfer of an ion present in dialysate such as sodium, the dialysance D must be used and it is given by the same equation as Equation 18.7. 18.2.5 Hemofiltration and Hemodiafiltration

Hemoﬁltration (HF), proposed initially by Funck-Brentano et al. [10] and Henderson et al. [11] consists in relying only upon the ultraﬁltration for toxin removal. There is no dialysate circuit and the ultraﬁltration (UF) ﬂow rate must be at least equal to 110 ml/min to produce enough urea clearance. This condition necessitates using highly permeable membranes and a large blood ﬂow, 300 ml/min or more in the dialyzer. This high UF ﬂow rate must be compensated by reinjection of sterile dialysate in the blood line, generally at the hemoﬁlter outlet, in order to prevent hypovolemia and the loss of plasma ions. This reinjection ﬂow rate is slightly below that of the UF ﬂow rate, in order to remove excess ﬂuid as in regular dialysis.

j417

j 18 Biomedical Membrane Extracorporeal Devices

418

The clearance can be easily calculated from Equation 18.4 and the toxin mass balance on the hemodialyzer, which is Q F CF ¼ Q Bi CBi Q Bo CBo

ð18:9Þ

Combining Equations 18.4 and 18.9 yields KHF ¼ Q F CF =CBi ¼ TrQ F

ð18:10Þ

since CF/CBi is the transmittance (Tr) of the membrane for the toxin considered. This transmittance is equal to 1 for small toxins such as urea and creatinin and decreases as the molecular weight of the toxin increases, but less rapidly than the diffusive clearance of Equation 18.7. Thus, hemoﬁltration produces a lower clearance of small toxins than hemodialysis, but a higher one for large toxins such as myoglobin and Beta2 microglobulin. Hemoﬁltration has been shown to improve vascular stability and prevent hypotension. But it is a more expensive treatment than regular dialysis, due to the higher cost of highly permeable membranes and of sterile bags of reinjection ﬂuid prepared in pharmaceutical plants. Presently, hemoﬁltration is mostly restricted to acute dialysis that takes place 24 h per day for one or two weeks and necessitates a smaller urea clearance than HD. Hemodiaﬁltration (HDF) may be regarded as a combination of regular HD and hemoﬁltration as it associates a dialysate circulation and high ultraﬁltration. It also necessitates the reinjection of sterile dialysate, but less than in hemoﬁltration, since the clearance of small toxins is also achieved by dialysis. This process, thus, combines the high urea clearance of dialysis with the high rate of large toxins removal of hemoﬁltration. In addition, generators now exist that can prepare simultaneously the regular dialysate from concentrate and the reinjection ﬂuid by sterile ﬁltration of this dialysate, therefore lowering the treatment cost. The overall clearance of hemoﬁltration is more difﬁcult to calculate than the diffusive clearance or the HF clearance, as it combines diffusive and convective transfers. An approximate equation for this clearance, obtained from an exact numerical solution has been given by Jaffrin et al. [12] as KHDF ¼ KD þ 0:43Q F þ 0:00083Q 2F

ð18:11aÞ

where Q F and K D are in ml/min, or if Q F < 70 ml/min KHDF ¼ KD þ 0:46Q F

ð18:11bÞ

which shows that the overall clearance is generally less than the sum of diffusive and convective clearance, due to interaction between diffusive and convective transfers. Equation 18.11b can also be applied to regular hemodialysis and includes the contribution of the convective transfer to the clearance. 18.2.6 Various Types of Hemodialyzers

The most common type is the hollow-ﬁber cartridge, consisting in a bundle of 10 000 to 15 000 ﬁbers of 200–220 mm i.d., placed in a polypropylene transparent housing.

18.2 Hemodialyzers

Figure 18.5 Cut out of a hollow-fiber dialyzer.

Blood circulates inside the ﬁbers and dialysate outside (Figure 18.5). The ﬁbers are imbedded at each end in polyurethane that ensures separation between blood and dialysate. During fabrication, the bundle ends must be carefully cut with a blade to reopen each ﬁber. Blood inlet and outlet are located in caps screwed on the housing, which are designed to distribute blood uniformly into all ﬁbers, while dialysate ports are located at each end of the housing, perpendicular to the ﬁbers. Parallel-plate hemodialyzers using ﬂat membranes, with several compartments in parallel, separated by plastic plates, are now only available from Hospal Co (Crystal and Hemospal models). Blood circulates between two membranes and the dialysate between the other side of membrane and the plastic plate. These parallel-plate dialyzers have a smaller blood-pressure drop than hollow-ﬁber ones and require less anticoagulants as ﬂat channels are less exposed to thrombus formation than ﬁbers, but they are heavier and bulkier and thus less popular. A recent survey of the stateof-the-art in hemodialyzers is given in [13]. 18.2.6.1 Various Types of Membranes There are two types of membranes, cellulosic and synthetic or polymeric ones. Cellulosic membranes can be in regenerated cellulose (cuprophan, Bioﬂux from Membrana, Germany) or modiﬁed cellulose (cellulose acetate or diacetate, from Asahi, triacetate cellulose from Baxter and Nipro, which has a high hydraulic permeability or Hemophan from Membrana). Cuprophan was originally the most common one, because of its low cost, but is no longer produced because of its lower biocompatibility and hydraulic permeability. A wide variety of polymeric membranes are now available with both high and medium hydraulic permeabilities. Only the Eval

j419

j 18 Biomedical Membrane Extracorporeal Devices

420

Figure 18.6 SEM pictures of hollow-fiber membranes used for HD and HDF.

(Kuraray) is naturally hydrophilic, but the other polymers can be made hydrophilic by incorporating additives during their fabrication. SEM pictures of various membranes are shown in Figure 18.6. The ﬁrst high-ﬂux membrane was the polyacrylonitrile AN69 introduced in 1973 by Hospal in France, followed by the PAN from Asahi, in Japan. These membranes, which are symmetric, were considered at this time as very biocompatible, but adsorbed proteins due to their negative charge. This adsorption was reduced by coating the AN69 with polyethylemine to reduce its negative charge in the AN69 ST. Polyamide asymmetric membranes (Gambro) were blended with PVP (polyvinylpyrrolidone) to make them hydrophilic. Polysulfone membranes introduced ﬁrst by Fresenius, but also available from Asahi and Toray in Japan, had high hydraulic permeability together with excellent biocompatibility. An improved version (Helixone) with a reduced inner diameter of 185 mm to increase clearance and a mean pore size increased from 3.1 to 3.3 nm was introduced by Fresenius in 2000 as a high-ﬂux membrane. Other synthetic membranes include the polymethylmethacrylate (PMMA, Toray), polycarbonate polyether (Gambro), and polyethersulfone (Membrana). 18.2.6.2 Optimization of Hemodialyzer Performance The goal is to improve clearance without increasing membrane area or dialysate ﬂow rate, as it would increase treatment cost. The dialysate diffusive resistance RD can be lowered by improving dialysate ﬂow distribution with multiﬁlament spacer yarns between ﬁbers or weaving these yearns around ﬁbers. Another approach consists in giving the ﬁbers a wavy shape by reducing the housing length below that of ﬁbers. It is important to ﬁnd the optimal packing density, which is the best compromise between a loose packing resulting in a low average dialysate velocity and a dense packing that creates preferential channels between regions of almost stagnant ﬂow. Decreasing the ﬁber inner diameter while increasing the number of ﬁbers to keep the total area constant will both decrease the blood diffusive resistance RB and increase diffusive clearance and also raise the blood pressure drop and the ultraﬁltration in the upstream part of the membrane [14]. In fact, it is possible by creating a high UF in the ﬁrst half of the ﬁlter and a back ﬁltration in the second half to simulate HDF conditions in a hemodialyzer without the need of reinjection ﬂuid. The ﬂuid

18.3 Plasma Separation and Purification by Membrane

loss by ultraﬁltration is compensated by the back ﬁltration of the incoming fresh dialysate, which contains only a small quantity of toxins.

18.3 Plasma Separation and Purification by Membrane 18.3.1 Introduction

In contrast to hemodialysis that uses ultraﬁltration membranes, plasma separation (also called plasmapheresis) requires microﬁltration membranes with a pore size from 0.2 to 0.6 mm, in order to transmit all proteins and lipids, including LDL cholesterol (2000 kDa) and retain completely platelets (2 mm diameter), red blood cells (8 mm diameter) and white blood cells. Thus, membrane plasmapheresis can yield high-quality platelet-free plasma and red cells can be either continuously returned to the donor or saved in another bag for blood transfusion. But it is important, in the case of plasma collection from donors, to minimize the membrane area, in order to reduce the cost of disposable hollow-ﬁber ﬁlters and to avoid the risk of hemolysis (free hemoglobin release) due to RBC damage by contact at the membrane if the pressure difference across the membrane is too high. Membrane plasmapheresis is also the ﬁrst step for treatment of pathological plasma in the case of autoimmune diseases, as the patient retains his own red blood cells while his plasma is replaced by an albumin solution or fresh frozen plasma obtained from donors (plasma exchange therapy). Other more selective plasma puriﬁcation techniques consist in eliminating pathologic immunoglobulins or LDL cholesterol familial hypercholesterolemia, either by a secondary ﬁltration, chemical adsorption or immunoadsorption. A description of various applications of plasmapheresis can be found in the book edited by Smit Sibinga and Kater [15]. In France alone, about 220 000 plasmapheresis and 65 000 cytapheresis (collection of platelets, factor VIII, etc.) are performed every year, against 2 400 000 blood donations. 600 ml of plasma can be collected from the same donor every 2 weeks if needed. 18.3.2 The Baxter Autopheresis C System for Plasma Collection from Donors

This system was ﬁrst introduced by Hemascience Company, Santa Ana, CA, USA. The ﬁlter consists of a cylindrical membrane of only 58 cm2 area rotating at 3600 rpm inside a concentric cylinder of 2.9 cm inner diameter (Figure 18.7). Blood inlet and outlet are mounted tangentially, and plasma is collected through grooves molded on the inner cylinder supporting the membrane and leaves the ﬁlter through a duct in the rotation axis at the bottom. The presence of Taylor vortices generated in the gap between the membrane and housing by the rotation, create very high shear rates at membrane, producing a large

j421

j 18 Biomedical Membrane Extracorporeal Devices

422

Figure 18.7 Schematic of rotating cylindrical filter for plasma collection from donors.

plasma permeate ﬂux about 0.5 cm/min or 300 L/hm2, which permits the use of a very small membrane area. This system was later commercialized by Baxter Company under the name Autopheresis C. Initially, the membrane was in polycarbonate with 0.8-mm pores, but, since its peak ﬂux occurred in a narrow transmembrane pressure range, it was replaced later by a nylon one (Pall Corp) with 0.5-mm pores, which yielded a constant permeate ﬂux above 50 mmHg. This device was successfully applied to therapeutic plasma exchange by Kaplan et al. [16], who, after increasing the pressure setting, were able to collect 3 L of plasma from a patient in 90 min. An interesting modiﬁcation to this device has consisted in separating ﬁrst by centrifugation the blood into platelet-rich plasma using a similar rotating system without a membrane. Then the platelets were separated from the plasma using the Autopheresis C, with minimal contamination by leukocytes and red cells. 18.3.3 Therapeutic Applications of Plasma Separation

Membranes used for therapeutic plasma separation have the same characteristics as those used for plasma collection from donors, but their area is larger as the amount

18.3 Plasma Separation and Purification by Membrane

Figure 18.8 Hollow-fiber plasma filter with its extracorporeal circuit for plasma exchange.

of plasma ﬁltered is bigger. They are generally made of synthetic hollow ﬁbers with inner diameter of 250 to 350 mm. Their area vary from 0.2 m2 to about 0.65 m2 for therapeutic plasmapheresis with a unit cost from D 30 to 250 depending upon their size. 18.3.3.1 Plasma Exchange Plasma exchange can be performed by membrane ﬁltration to avoid RBC damage, which centrifugation cannot guarantee. The removed plasma (2.5–3.5 L) is discarded and replaced by fresh frozen plasma collected from donors or albumin concentrate (Figure 18.8). Table 18.3 lists available plasmapheresis ﬁlters and their characteristics. This technique is applied in various specialties such as nephrology, haematology and neurology, such as thrombotic thrombocytopenic purpura, myasthenia gravis, chronic inﬂammatory demyelinating polyneuropathy, Waldenstrom macroglobulinemia, the Guillain-Barre syndrome, rheumatoid arthritis, systemic lupus erythematosus, and multiple sclerosis. 18.3.3.2 Selective Plasma Purification by Cascade Filtration The treatment of several autoimmune diseases and of familial hypercholesterolemia generally requires the removal of a single pathogeneous molecule such as LDL, IgA or IgG for Guillain-Barre syndrome, IgM for Walderstom macroglobulinemia, and so on. Removing speciﬁc molecules avoids the need of substitution by albumin or fresh frozen plasma (FFP) to compensate protein loss as in case of total plasma exchange. Cascade ﬁltration, which was initially proposed by Agishi et al. [17], consists in ﬁltering the collected plasma after separation on an ultraﬁltration ﬁlter selected so as to retain the pathogeneous molecule, as shown in Figure 18.9. This technique is now

j423

j 18 Biomedical Membrane Extracorporeal Devices

424

Table 18.3 Commercially available membrane filters for plasma separation and fractionation.

Manufacturer Plasma separators Baxter, USA Gambro, Sweden Fresenius, Germany Dideco, Italy Asahi, Japan Kuraray, Japan Toray, Japan Terumo, Japan Nippro, Japan Gambro

Filter

Membrane

Membrane area (m2)

Pore diam. In fiber (lm) diam. (lm)

CPS-10 PP Plasmaﬂux

Polypropylene Polypropylene Polypropylene

0.17 0.38 0.50

0.55 0.5 0.5

Hemaplex Plasmaﬂo 03-06-08 Plasmacure Plasmax PS 02-05 PS-4000 PEX-50 Prisma TPE

Polypropylene Polyethylene

0.20 0.5 0.3–0.6–0.8 0.2

320 370

Polysulfone PMMA

0.30 0.150–0.5

0.2 0.5

300 370

Cellulose acetate Cellulose triacetate Polypropylene

0.50 0.50 0.35

0.45 0.4 0.5

80 (height)a 270 330

1.00

0.02

350

1.7

106b

210

1.7

5 · 106b

210

1.00

0.01–0.02

200

0.80

c

370

Plasma fractionators (secondary membranes) Dideco, Italy Albusave Cellulose diacetate BT902 Asahi, Japan Cascadeﬂo Cellulose diacetate AC1730 Cascadeﬂo Cellulose diacetate AC1760 Kuraray, Japan Eval ﬁlter Ethylene vinyl 2A-3A alcohol Toray, Japan Plasmax PMMA AS08

320 330 330

Channel height (plate-type ﬁlter). Membrane cutoff at 90% rejection. c Cutoff not available. a b

mostly used for removal of LDL (bad cholesterol) as its molecular weight (2000 kDa) is much larger than that of HDL (good cholesterol, 400 kDa) albumin (69 kDa) and IgG (156 kDa), which are easily transmitted by the secondary membrane. This second ﬁltration is generally carried out in quasi-dead-end mode with a retentate ﬂow rate which is less than 15% of inlet ﬂow rate, in order to minimize albumin losses in the discarded retentate. However, like all ﬁltration methods, it is subject to membrane fouling, which decreases permeate ﬂux and selectivity during the ﬁltration and operating conditions need to be optimized in order to maximize selectivity. Thermoﬁltration consists in warming the plasma to 42 C to prevent cryogel formation in the secondary ﬁlter [18]. Table 18.3 lists available ﬁlters for plasma puriﬁcation or fractionation. Their membrane area is larger than that of plasma ﬁlters in order to reduce fouling and they are generally more expensive.

18.3 Plasma Separation and Purification by Membrane

Figure 18.9 Schematic of a cascade filtration circuit for plasma purification with secondary membrane.

A review of plasma puriﬁcation using secondary ﬁltration has been presented by Siami et al. [19] and Table 18.4 lists the diseases treated with this technique. Diseases treated include immune-mediated disorders and familial type IIA hypercholesterolemia. These authors concluded that cryoglobulins ﬁlters were safe and effective for removing cryoproteins, did not induce complement activation and constituted one of the most promising techniques of secondary membrane application.

Table 18.4 Diseases treated with secondary membrane filtration (From Siami et al. [8]), with permission.)

Disease

Pathogen removed

Membrane type

Myasthenia Gravis Guillain Barre Syndrome IgG (IgA) Gammopathy Idiopathic throbocytopepenic Purpura Factor XI deﬁciency Type I Cryoglobulinemia Waldenstrom Macroglobulinemia Castleman Syndrome Familial Hypercholesterolemia Type II Cryoglobulinemia Type III Cryoglobulinemia

Anti-AChRab IgG antibody IgG (IgA) antibody IgG platelet antibody FXI IgG inhibitor Monoclonal IgG IgM antibody IgM antibody LDL-Cholesterol Mono IgM and poly IgG Polyclonal IgM or IgG

2A Evaﬂux 2A Evaﬂux 2A Evaﬂux 2A Evaﬂux 2A Evaﬂux 2A Evaﬂux 4A Evaﬂux 4A Evaﬂux 4A Evaﬂux Cryoglobulin ﬁlter Cryoglobulin ﬁlter

j425

j 18 Biomedical Membrane Extracorporeal Devices

426

18.4 Artificial Liver 18.4.1 Introduction

The adult human liver normally weighs between 1.7–3.0 kg. It is both the second largest organ and the largest gland within the human body. The liver performs many important metabolic functions: detoxiﬁcation, transformation, storage and synthesis (Figure 18.10). Loss of liver cell functions resulting in the disruption of many essential functions could lead to death. At present, transplantation is the only efﬁcient treatment for patients suffering from acute or fulminant liver failure [20]. The shortage in speciﬁc organ donors has resulted in a high death rate among the potential patients waiting for a graft. for the past 20 years, the expanding gap between the number of patients on the waiting list and the number of liver transplants has led to the design of temporary liver support. Such an artiﬁcial organ could be employed either as a bridge to transplantation or as a means for the patient to recover native liver function [21]. As liver performs multiple and complex functions, artiﬁcial organ or bioartiﬁcial organ exploiting a synthetic cartridge to host biological components such as cells (hepatocytes in the case of a bioartiﬁcial liver) have been investigated. Among all of these potentialities, we only focus here on purely artiﬁcial systems. Membrane-based bioartiﬁcial livers (BAL) will not be described here, but could be found in other reviews [22–25] and in another chapter in the present book. 18.4.2 Physical Principles

One of the major liver functions is detoxiﬁcation of substances carried by blood, which are perfused through the cellular network in the organ. These functions can

Figure 18.10 Classification of major liver functions, with some examples.

18.4 Artificial Liver

Figure 18.11 Generic representation of combined filtration and adsorption columns systems for artificial liver support.

theoretically be replaced by the three different physical principles available in extracorporeal devices (Figure 18.1), alone or in association. Artiﬁcial liver support systems aim at the extracorporeal removal of water soluble and protein-bound toxins (albumin being the preferential binding protein) associated with hepatic failure. Albumin contains reversible binding sites for substances such as fatty acids, hormones, enzymes, dyes, trace metals and drugs [26] and therefore helps elimination by kidneys of substances that are toxic in the unbound state. It should be noticed that the range of substances to be removed is broad and not completely identiﬁed. Clinical studies showed that the critical issue of the clinical syndrome in liver failure is the accumulation of toxins not cleared by the failing liver. Based on this hypothesis, the removal of lipophilic, albumin-bound substances, such as bilirubin, bile acids, metabolites of aromatic amino acids, medium-chain fatty acids, and cytokines, should be beneﬁcial to the clinical course of a patient in liver failure. For this purpose, the removal procedures are mainly based on membrane separation that ideally should bring free and bound toxins to a nonspeciﬁc adsorption device (ion-exchangers and/or activated charcoal). Blood should not perfuse directly such components, due to bioincompatibity aspects. Therefore, several processes have been proposed to correctly handle toxins carried by plasma [27]. They are described in the following sections. All of them need a physical barrier between the blood cells and the adsorption system. This physical sieve is always a membrane with adequate properties, through which toxins can be transferred by diffusion or convection. Two systems are then based on classical chromatography processes, that is, perfusion of a column hosting adsorbents, and three others are based on moving adsorbent phase.

j427

j 18 Biomedical Membrane Extracorporeal Devices

428

18.4.3 Convection þ Adsorption Systems

The following processes can be described as selective therapeutic plasmapheresis. In a ﬁrst step, blood is withdrawn from the patient and separated by crossﬂow ﬁltration in a hollow-ﬁber membrane cartridge: water and some plasma solutes are transferred through a semipermeable membrane under a convection process. The transmembrane pressure applied from blood to ﬁltrate compartment ensures ﬂow and mass transfers. Then, the ﬁltrate perfuses the adsorption columns where toxins are retained and is ﬁnally mixed with blood cells and other plasma components before returning to the patient (Figure 18.11). In the ASAHI KASEI Medical (Tokyo, Japan) system, the plasmapheresis step is performed by a microporous membrane (Plasmaﬂo) made of a copolymer of ethylene and vinyl alcohol (PEVA), with a maximum pore size of 0.3 mm. The extracted plasma ﬂows through an activated charcoal column Hemosorba and an anion-exchange column (copolymer of styrenedivinyl benzene) Plasorba that binds bilirubin and bile acids [28]. Each column contains 350 mL of adsorbent. In the Prometheus (Fresenius Medical Care, Bad Homburg, Germany) system, blood fractionation is achieved by means of a capillary albumin ﬁlter (AlbuFlow AF01) presenting a molecular weight cutoff of 300 000 Da. The large pores of the membrane allow albumin-rich plasma to enter the secondary plasma circuit. Albumin-bound toxins are extracted from plasma after binding to the adsorber beads. Both columns host 350 mL of neutral resin styrenedivinyl benzene copolymer beads. The inner porous structure is sponge-like and easily accessible for protein-bound liver toxins. Prometh-01 adsorbs water-insoluble compounds, such as bile acids, phenols and aromatic amino acids. Prometh-02 has anion-exchanger properties because it contains positively charged sites and is able to remove negatively charged liver toxins, such as bilirubin. The cleared ﬁltrate returns thus to the blood main stream [29]. To remove water-soluble toxins, an additional classical dialysis step is then placed downstream. The high-ﬂux dialysis membrane is used for the diffusive transfer of toxins from the blood to the dialysate side. The whole extracorporeal circuit is adapted from a Fresenius 4008 dialysis machine [30]. 18.4.4 Diffusion þ Adsorption Systems

In these cases, substances carried by blood are removed in the dialysate phase, which is separated from blood by a semipermeable membrane, as described in Figure 18.2. Toxins should cross this barrier by diffusion, before being treated. Toxins that bind to albumin have proven refractory to removal by conventional hemodialysis. Such toxins can, however, be removed by adding binders to the dialysate that capture them after being dialyzed through the membrane. In the Liver Dialysis Unit (HemoCleanse Technologies, Lafayette, USA), the adsorbents (powdered activated charcoal and cation-exchangers) are located in the dialysate-moving phase [31]. Dialysate content is adjusted so as to prevent

18.4 Artificial Liver

unexpected removal of substances such as calcium, sodium . . . due to diffusive effects [32]. The MARS system from Teraklin (a subsidiary of Gambro, Lund, Sweden) uses a specially designed membrane named MARS Flux dialyzer in the primary circuit. The secondary circuit consists of 600 mL of 20% human albumin solution instead of physiological buffer used in classical hemodialysis. Due to the membrane properties, protein-bound toxins and water-soluble substances from the blood side diffuse into the albumin solution [33]. The MARS Flux dialyzer presents a surface area of 2.1 m2, a membrane thickness of 100 nm and a molecular cutoff of about 50 kDa. The speciﬁc membrane surface provides pseudobinding sites for albumin when the secondary circuit is primed with albumin solution. The albumin molecules on the dialysate side of the membrane are in very close proximity to the surface of the membrane in contact with patients blood. It is assumed that albumin-bound toxins move by physicochemical interactions between the plasma, albumin molecules bound to the dialysis side of the membrane and the circulating albumin solution. This solution is then dialyzed against a standard buffered dialysis solution to remove water-soluble substances by diffusion. The removal of the albumin-bound toxins is achieved by an activated-carbon adsorber and an anion exchanger. The concentration gradient is maintained at the ﬁrst dialysis step by circulation of the online regenerated albumin [34]. A speciﬁc monitor has been designed by Teraklin Company to handle the secondary circuit and is now commercially available with Gambro dialysis machines. The single-pass albumin dialysis (SPAD) is a simple noncommercial method of albumin dialysis using standard renal-replacement therapy machines without an additional perfusion pump system. The patients blood ﬂows through a circuit with a high-ﬂux hollow-ﬁber hemodiaﬁlter, identical to that used in the MARS system. The other side of this membrane is perfused with a buffered albumin solution in counterdirectional ﬂow, which is, instead of being regenerated as in the MARS concept, discarded after passing the ﬁlter [35]. 18.4.5 Future of Artificial Livers

As seen above, the artiﬁcial systems are only able to supply detoxication functions of the liver. In some cases, this might not be enough to save patients. An alternative is the design of bioartiﬁcial liver. A simplistic approach consists in considering such a device as a bioreactor based on synthetic elements able to offer an adequate environment to the liver cells. This environment would in turn lead to the maintenance of efﬁcient functions of the cells aiming at liver supply, when placed in a bioreactor located in an extracorporeal circuit. The mandatory requirements for acceptable cell viability and functions in a bioartiﬁcial liver (BAL) are tentatively listed below, according to a biotechnological point of view: (a) Anchorage to a support or a matrix; (b) Effective exchanges with blood or plasma; (c) Protection from host immunological response.

j429

j 18 Biomedical Membrane Extracorporeal Devices

430

18.4.6 Conclusions

Up to now, none of the presented system can claim its ability to fully replace all liver functions in an extracorporeal circuit. On the one hand, purely artiﬁcial techniques can only cover some detoxiﬁcation aspects, which is already crucial in many clinical cases to save patients. On the other hand, bioartiﬁcial livers have not proven their full efﬁciency yet, mainly because both regulatory and logistic aspects limit, for the moment, the inclusion of signiﬁcant numbers of patients to draw statistically relevant conclusions. It seems, nevertheless, clear that the combination of membrane-based and adsorbent techniques, perhaps in addition to bioartiﬁcial systems, present a potential supply to help the patient waiting for a graft or even for tissue regeneration. In the biomedical ﬁeld, the extension of techniques previously developed for other topics, such as biochromatography for instance, has always proved to be promising. This could hopefully be the case for artiﬁcial liver support.

References 1 Solomon, B.A., Castino, F., Lysaght, M.J., Colton, C.K. and Friedman, L.I. (1978) Transactions of the American Society for Artiﬁcial Internal Organs, 24, 21–26. 2 Samtleben, W., Hillebrand, G., Krummer, D. et al. (1980) in Plasma Exchange (ed. H.G. Sierbeth), Schattauer, Stuttgart, pp. 175–178. 3 Nose, Y., Mikami, J. and Kasai, S. (1963) Transactions of the American Society for Artiﬁcial Internal Organs, 9 (358), 362. 4 Opolon, P., Rabin, J.-R., Huquet, C., Granger, A., Delorme, J.L., Boschat, M. and Sausse, A. (1976) Transactions of the American Society for Artiﬁcial Internal Organs, 22, 701–710. 5 (2006) ESRD Patients in 2005: A Global Perspective. Fresenius Medical Care document. 6 Kunimoto, T., Lowrie, E.G., Kumazawa, S., OBrien, M., Lazarus, J.M., Gottlieb, J.M., Merril, J.P. (1977) Transactions of American Society of Artiﬁcial Internal Organs, 23, 234–242. 7 Ahrenholz, P., Winkler, R.E., Ramlow, W., Tiess, M. and Mueller, W. (1997) The

8

9

10

11

12

13

International Journal of Artiﬁcial Organs, 20, 81–90. Sargent, J.A. and Gotch, F.A. (1978) Principles and biophysics of dialysis, in Replacement of Renal Function by Dialysis (eds W. Von Drucker, F.M. Parsons and R.F. Maher), Nijhoff, Boston, pp. 38–68. Lysaght, M.J., Ford, C.A., Colton, C.K. et al. (1977) Mass transfer in clinical blood ultraﬁltration devices, in Technical Aspects of Renal Dialysis (ed. T.M. Frost), Pitman Medical, London, UK. Funck-Brentano, J.L., Sausse, A., Man, N.K. et al. (1972) Proceedings of European Dialysis and Transplant Association Congress, 9, 52–58. Henderson, L.W., Colton, C.K. and Ford, C.A. (1975) The Journal of Laboratory and Clinical Medicine, 85, 372–390. Jaffrin, M.Y., Ding, L.H. and Laurent, J.M. (1991) Journal of Biomechanical Engineering-Transactions of the ASME, 112, 212–219. Uhlenbusch-Korwer, I., Bonnie-Schorn, E., Grassmann, A. and Vienken, J. (2004) Understanding Membranes and Dialyzers, Pabst, Berlin.

References 14 Ronco, C., Brendolan, A., Lupi, A., Metry, G. and Levin, N.W. (2000) Kidney International, 58, 809–817. 15 Smit Sibinga, C.T. and Kater, L. (eds) (1991) Advances in Haemapheresis, Kluwer, Amsterdam. 16 Kaplan, A.E. and Halley, S.E. (1988) Transactions of the American Society for Artiﬁcial Internal Organs, 34, 274–276. 17 Gurland, H.J., Samtleben, W. and Blumenstein, M. (1983) Life Support Systems, 1, 61–70. 18 Agishi, T., Kaneko, I. Hasuo, Y.et al. (1980) Transactions of the American Society for Artiﬁcial Internal Organs, 26, 406–410. 19 Siami, F.S. and Siami, G.A. (2000) ASAIO Journal (American Society for Artiﬁcial Internal Organs: 1992), 46, 383–388. 20 Chapman, R.W., Forman, D., Peto, R. and Smallwood, S. (1990) Lancet, 335, 32. 21 Cao, S., Esquivel, C.O. and Keeffe, E.B. (1998) Annual Review of Medicine, 49, 85. 22 Arkadopoulos, N., Detry, O., Rozga, J. and Demetriou, A.A. (1998) The International Journal of Artiﬁcial Organs, 21, 781. 23 Legallais, C., David, B. and Dore, E. (2001) Journal of Membrane Science, 181, 81. 24 Matsushita, M. and Nose, Y. (1986) Artiﬁcial Organs, 10, 378. 25 Nyberg, S.L. and Misra, S.P. (1998) Mayo Clinic Proceedings, 73 (8), 765. 26 Emerson, T.E. (1989) Critical Care Medicine, 17, 690. 27 Legallais, C., Vijayalakshmi, M.A. and Moriniere, P. (2002) Biochromatography

28

29

30

31

32

33

34

35

and Biomedical Applications, in Biochromatography Theory and Practice (ed. M.A. Vijayalakshmi), Taylor and Francis, London and New York, pp. 496. Iseki, J., Touyama, K., Nakagami, K., Takagi, M., Hakamada, K., Ooba, N. and Mori, N. (1995) Hepatogastroenter, 42, 394. Vienken, J. and Christmann, H. (2006) Therapeutic Apheresis and Dialysis, 10, 125. Santoro, A., Faenza, S., Mancini, E., Ferramosca, E., Grammatico, F., Zucchelli, A., Facchini, M.G. and Pinna, A.D. (2006) Transplantation Proceedings, 38, 1078. Ash, S.R., Knab, W.R., Blake, D.E., Carr, D.J., Steczko, J., Harker, K.D. and Levy, H. (2000) Therapeutic Apheresis, 4 (3), 218. Bauer, E., Gendo, A., Madl, C., Garo, F., Roth, E. and Kramer, L. (2002) The International Journal of Artiﬁcial Organs, 25, 923. Stange, J. and Mitzner, S. (1996) The International Journal of Artiﬁcial Organs, 19, 677. Klammt, S., Stange, J., Mitzner, S.R., Peszynski, P., Peters, E. and Liebe, S. (2002) Liver, 22, 30. Sauer, I.M., Goetz, M., Steffen, I., Walter, G., Kehr, D.C., Schwartlander, R., Hwang, Y.J., Pascher, A., Gerlach, J.C. and Neuhaus, P. (2004) Hepatology (Baltimore, MD), 39, 1408.

j431

j433

19 Membranes in Regenerative Medicine and Tissue Engineering Sabrina Morelli, Simona Salerno, Antonella Piscioneri, Maria Rende, Carla Campana, Enrico Drioli, and Loredana De Bartolo

19.1 Introduction

The ﬁelds of tissue engineering and regenerative medicine aim at promoting the regeneration of tissues or replacing failing or malfunctioning organs, by means of combining a scaffold/support material, adequate cells, and bioactive molecules. The use of materials in contact with biological materials (cells, tissues/organs, physiological ﬂuids, and biomolecules) is a current illustration of the need for interdisciplinary scientiﬁc approaches that combine the most recent advances in materials science and technology, basic sciences and life sciences. Different materials have been proposed to develop membranes to support cells and promote their differentiation and proliferation towards the formation of a new tissue. Such strategies allow for producing hybrid constructs that can be implanted in patients to induce the regeneration of tissues or replace failing or malfunctioning organs. In recent years rapid progress has been made in the ﬁeld of biomedical materials that utilize both natural and synthetic polymers and that can be used in a variety of applications, including wound closure, drug-delivery systems, novel vascular grafts, or scaffolds for in-vitro or in-vivo tissue engineering. The goal of the early or ﬁrstgeneration biomedical materials, during the 1960s and 1970s, was to attain suitable physical properties to match the replaced tissue with a common feature of biological inertness. Second-generation biomaterials were designed to produce bioactive responses that could elicit a controlled reaction in the physiological environment. Such bioactive (ceramics, hydroxyapatite) or resorbable (polyglycolide, polylactide) materials have been successfully applied to the medical needs of many ﬁelds. Thirdgeneration biomaterials are combining these two properties and are being designed to stimulate speciﬁc cellular responses at the molecular level [1]. Several smart biomaterials for tissue engineering and regeneration are activated by either cells or genes and are designed to improve the complicated biological event of tissue repair. It was demonstrated that among polymeric materials, polymeric membranes are attractive for their selectivity and biostability characteristics in the use of biohybrid

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 19 Membranes in Regenerative Medicine and Tissue Engineering

434

systems for cell culture. Semipermeable membranes act as a support for the adhesion of anchorage-dependent cells and allow the speciﬁc transport of metabolites and nutrients to cells and the removal of catabolites and speciﬁc products [2–4]. Moreover, new membrane systems that have been recently realized might also potentially contribute to regenerative medicine and tissue engineering. Membranes are used as selective barriers to entrap living cells to be transplanted avoiding contact with immunocompetent species [5, 6] or in dynamic systems used as organ typical system. New strategies in the development of new systems might lead to membranes that are able to stimulate speciﬁc cell responses and maintain differentiated functions. In addition, the growing interest in synthetic and biodegradable polymeric biomaterials for tissue engineering and human cell therapies has led to novel approaches to improve cell–biomaterial interactions: the development of new biocompatible and cytocompatible materials and modiﬁcation of surface chemistry including grafting of functional groups, peptides and proteins leaving the bulk properties unaltered. For example, the incorporation of a signal peptide such as RGD (Arg-Gly-Asp) into the biomaterial was an attempt to mimic the extracellular matrix, modulate cell adhesion, and induce cell migration. An intermediate density of adhesive ligand is crucial for optimal cell migration. Cell-speciﬁc recognition factors can be incorporated into the resorbable polymer surface, including the adhesive protein ﬁbronectin or functional domains of ECM components [7]. The polymer surface can be tailored with proteins that inﬂuence interactions with endothelium [8], synaptic development [9], and neurite stimulation [10]. On the basis of these important considerations it can be pointed out that the development of new biomaterials able to activate a speciﬁc response of the cells and to maintain cell differentiation for a long time is one of the most pertinent issues in the ﬁeld of tissue engineering and regenerative medicine. Nowadays, it has been demonstrated that the suitability of polymers for tissue-engineering purposes is highly dependent on the tissue that needs to be engineered. The histological, physiological, and biomechanical properties of each tissue determine the success of the regenerative process, therefore restricting the choice of materials. This paper reports on the development of membrane systems to be used for cell culture (e.g., hepatocytes, lymphocytes, neuronal cells) in biohybrid systems such as a therapeutic device or as in-vitro model systems for studying the effects of various drugs and chemicals on cell metabolism.

19.2 Membranes for Human Liver Reconstruction

Liver failure is a fatal disease and liver transplantation is the only established treatment; however, donor shortages remain problematic. Liver failure is potentially reversible because of liver regeneration [11] so considerable work has been done over many years to develop effective liver-support devices and various hepatic support systems using hepatocytes have been developed. Liver-engineered constructs can be also applied to the development of drugs to treat many diseases. The impact will be

19.2 Membranes for Human Liver Reconstruction

increasing for the coming decade in the design of in-vitro physiological models to study disease pathogenesis and in the development of molecular therapeutics. Since isolated hepatocytes may be able to undertake the full range of known in-vivo biotransformation and liver-speciﬁc functions [12], they could be used in vitro as model system for metabolic study. Hepatocytes are anchorage-dependent cells that require adhesive substrates for their functional and phenotypic maintenance [13]. So, a lot of artiﬁcial substrates such as membranes, microcarriers and biological matrix were studied for the hepatocyte cultures. During the last few years our experience has contributed to demonstrate that semipermeable synthetic membranes, owing to their structural and physicochemical characteristics, can be used for the development of biohybrid systems for cell cultures. We demonstrated that the hydrophobic/ hydrophilic properties of membranes, such as surface free-energy parameters, affect cell adhesion, cell morphology and speciﬁc metabolic function of hepatocytes [3, 14]. Previous studies showed that polymeric membranes are able to support the longterm maintenance of metabolic and biotransformation functions of isolated human hepatocytes in a biohybrid system [15, 16]. We prepared membranes from a polymeric blend of modiﬁed polyetheretherketone or PEEK-WC and polyurethane (PU) by an inverse phase technique by using the direct immersion-precipitation method. The developed PEEK-WC-PU membrane combines the advantageous properties of both polymers (i.e. biocompatibility, thermal and mechanical resistance, elasticity) with those of membranes such as permeability, selectivity and welldeﬁned geometry. This membrane is able to support cell adhesion and differentiation in a biohybrid system constituted of human hepatocytes and PEEK-WC-PU membrane for more than 1 month. A confocal microscopy image (Figure 19.1) evidences the 3D structure of hepatocytes after days of culture on a PEEK-WC-PU membrane; cells organize in small aggregates, which would lead to better functional maintenance, maintaining a polygonal shape, so many of the features of the liver in vivo are reconstituted.

Figure 19.1 Confocal images of human hepatocytes on PEEK-WCPU membranes by actin staining with FITC-phalloidin (green) and by nucleic acid staining with DAPI (blue). Scale bar 10 mm.

j435

j 19 Membranes in Regenerative Medicine and Tissue Engineering

436

In particular, the microﬁlaments of actin, which is the component of the cytoskeletal system that allows movement of cells and cellular processes, are localized along the cell periphery. In addition, there are junctional complexes within the aggregates that are linked to the cortical actin. This spatial actin ﬁlament distribution is similar to that observed with hepatocytes cultured in collagen, where cells assume a cuboidal shape with extensive cell apposition. The development of a structural conﬁguration resembling the native hepatic tissue involves the development of speciﬁc intercellular adhesions and redistribution of cell–cell and cell–surface adhesion forces, which are intimately related to the dynamic cytoskeletal organization. In human hepatocytes cultured in the membrane biohybrid system the localization of actin demonstrated the formation of cell–cell contacts that should provide better conditions for the maintenance of liver-speciﬁc functions. No spreading of cells was observed as a result of controlled interaction with the membrane surface. This study demonstrates that the PEEK-WC-PU membrane system is able to promote attachment and aggregate formation of cells outside of the body, providing a microenvironment able to elicit speciﬁc cellular responses of tissue analogues. In the last few years the use of membrane biohybrid systems has contributed to giving important information about the effect of various drugs, such as diclofenac, rofecoxib, paracetamol the effects of which on the speciﬁc functions of human hepatocyte are not completely known [17–21]. An interesting approach to the design of membranes able to activate speciﬁc biological responses of the cells is the surface-modiﬁcation technique by plasma process and biomolecule immobilization. This allows the cytocompatibility of membranes to be enhanced, leaving the bulk properties unaltered. It has been demonstrated that the immobilization of biomolecules on the membrane surface improves cell adhesion and the maintenance of differentiated functions [17, 21–24]. In particular, the RGD amino acid sequence (arginine–glycine–aspartic acid) stimulates cell adhesion on synthetic surfaces, since this oligopeptide represents the minimal adhesion domain of the majority of extracellular matrix (ECM) proteins (e.g., ﬁbronectin, vitronectin, and collagen) [17–22]. In the case of hepatocytes the immobilization of galactose motifs on the surface enhances the speciﬁc interaction with cells owing to the speciﬁc binding between the galactose moiety and the asyaloglycoprotein receptor present on the cytoplasmatic membrane [25, 26]. We modiﬁed a polyethersulfone membrane (PES) surface with a plasma-deposited acrylic acid coating (PES-pdAA) and RGD peptide covalently immobilized through a spacer arm molecule (SA), obtaining PES-pdAA-SA-RGD membranes. The same method was used to immobilize galactose in its acid form (Galactonic acid) to obtain the PES-PdAA-SA-GAL membranes. Figure 19.2 shows the different physicochemical properties of the PES membranes modiﬁed with RGD and galactose moiety with respect to the collagen that was used as a natural substrate. Native PES membranes have a very high hydrophilic surface character, in fact, the water contact angle measured on this membrane was 30 1.4 . Also, the modiﬁed membranes display a marked wettability even if, in this case, the

19.2 Membranes for Human Liver Reconstruction

Figure 19.2 Water contact angle of unmodified, modified membrane and collagen at t ¼ 0. PES-pdAA-SA-GAL and PESpdAA-SA-RGD are abbreviated to PES-GAL and PES-RGD, respectively. The reported values are the mean of 30 measurements of different droplets on different surface regions of each sample standard deviation.

galactose and RGD immobilization induced a reduction in the surface hydrophilic character: the values were 60 2.3 and 58 3.6 for PES-pdAA-SA-GAL and PESpdAA-SA-RGD membrane, respectively. Collagen ﬁlm displayed a higher contact angle with respect to the membranes (Figure 19.2). The performance of modiﬁed and unmodiﬁed membranes evaluated by analyzing the expression of liver-speciﬁc functions in terms of albumin production is shown in Figure 19.3. Hepatocytes cultured on unmodiﬁed PES membranes produced

Figure 19.3 Albumin production of human hepatocytes cultured on different modified membrane surfaces. PES-pdAA-SA-GAL and PES-pdA-SA-RGD are abbreviated to PES-GAL and PES-RGD, respectively. The values are the mean of six experiments standard deviation.

j437

j 19 Membranes in Regenerative Medicine and Tissue Engineering

438

albumin reaching values of 7.3 ng/ml h, while on modiﬁed membranes an increase of albumin production was observed. The highest levels of albumin synthesis were reached when hepatocytes were cultured on PES-pdAA-SA-GAL membranes where metabolic values of 11.9 ng/ml h were obtained. The ability of hepatocytes to synthesize albumin was also expressed at high level on PES-pdAA-SA-RGD membranes (10.8 ng/ml h). The immobilization of biomolecules, in particular with galactose moiety, improved the liver-speciﬁc functions of the hepatocytes cultured on their surface. It is interesting to note that the controlled molecular architecture of the membrane surface has a considerable impact on the liver cellular responses. A strategy that has been developed to ameliorate the in-vitro retention of hepatocyte functions includes the development of bioreactors using different materials, conﬁguration, and size [27–35]. Among these systems membrane bioreactors are particularly attractive because membranes allow the selective transport of metabolites and nutrients to cells and the removal of catabolites and speciﬁc products from cells [36, 37]. They also play the role of mechanical and chemical support for adhesion and growth of cells. In a previous study, human hepatocytes were cultured in a galactosylated polyethersufone (PES) membrane bioreactor, which permits cells to be cultured in a well-deﬁned ﬂuid dynamics microenvironment and metabolic rates to be easily estimated. Cells in the membrane bioreactor displayed an enhanced metabolic activity, which was maintained in the culture time at signiﬁcantly higher levels with respect to the batch system [22]. In particular, the cell metabolic functions of urea synthesis and protein secretion were maintained for 21 days (Figure 19.4). As reported in Figure 19.4 the urea synthesized by human hepatocytes reached values ranging from 28 to 60 mg/ml. Interestingly, the ability of cells to secrete proteins was also maintained for the whole period of culture demonstrating the good

Figure 19.4 Protein secretion (shade bars) and urea synthesis (full bars) of human hepatocytes cultured in the galactosylated membrane bioreactor for 21 days. The values are the mean of six experiments standard deviation.

19.3 Human Lymphocyte Membrane Bioreactor

performance of the bioreactor in the long-term maintenance of differentiated functions of cells outside of the body. This human hepatocyte galactosylated membrane bioreactor was also used as a model system to explore the modulation of the effects of a proinﬂammatory cytokine, Interleukin-6 (IL-6) on the liver cells at the molecular level [38] and gave evidence, for the ﬁrst time, that IL-6 downregulated the gene expression and synthesis of fetuin-A by primary human hepatocytes. The human hepatocyte bioreactor behaves like the in-vivo liver, reproducing the same hepatic acute-phase response that occurs during the inﬂammation process. This hepatocyte bioreactor should ﬁnd applications in drug testing, toxicological studies and in tissue engineering to help solve problems related to human diseases.

19.3 Human Lymphocyte Membrane Bioreactor

Isolated human lymphocytes are used to investigate the role of these cells in the pathogenesis of various diseases and in the autologous adoptive immunotherapy for patients with malignant disease or viral infections [39, 40]. Lymphocytes may be used also as biomarkers of target-organ susceptibility or as a marker of chemical effects and in the prediction of individual drug sensitivity alternatively to human liver biopsies [41, 42]. Generally, lymphocytes are cultured in static culture systems like T-ﬂasks, culture bags, well plates, which imply disadvantages in nonuniform culture conditions, low cell densities and uncontrolled process parameters [43–45]. This results in the maintenance of cell viability and functions for only a short time. Differently from static culture methods, dynamic systems such as bioreactors allow the culture of cells under tissue-speciﬁc mechanical forces such as pressure, shear stress, and interstitial ﬂow [46]. Among the various bioreactors (hollow ﬁber, stirred vessels, suspension bioreactor) [47–50] that have been developed to foster the retention of human lymphocytes in vitro one of the most used is the hollow-ﬁber (HF) membrane bioreactor. This bioreactor meets the main requirements for cell culture: a wide area for the exchange of oxygen/carbon dioxide and nutrient transfer, removal of catabolites and protection from shear stress. To date, the HF membrane bioreactor has mainly been used for large-scale mammalian cell culture to produce such products as monoclonal antibodies [48], for the expansion of tumor-inﬁltrating T lymphocytes [49]. Lamers et al. [47] demonstrated the usefulness of the HF bioreactor culture to produce cytokines by T lymphocytes. We developed a PEEK-WC hollow-ﬁber (HF) membrane bioreactor for the maintenance of human peripheral lymphocytes as a model system for the in-vitro investigation of disease pathogenesis, chemical effects and individual drug sensitivity. Peripheral lymphocytes isolated from the donors human buffy coat were cultured in the shell compartment of the PEEK-WC-HF bioreactor. Lymphocytes in the PEEK-WC-HF membrane bioreactor produced IL-2 and IL-10 throughout the culture period of 14 days (Figure 19.5).

j439

j 19 Membranes in Regenerative Medicine and Tissue Engineering

440

Figure 19.5 Cytokine production in human lymphocytes cultured in PEEK-WC-HF bioreactor for 14 days: (shade bars) IL-2 and (full bars) IL-10. The values are the mean of six experiments standard deviation.

IL-2 is the speciﬁc pattern of lymphocytes T helper 1 that increases the proliferation of T cells and activates B cells. IL-10 is a speciﬁc pattern of lymphocytes T helper 2, which is stimulatory towards certain T cells, mast cells and B cells and inhibits the synthesis of other cytokines produced by Th1. IL-10 was synthesized by cells with higher metabolic rates with respect to IL-2, suggesting a prevalent differentiation in lymphocyte immunophenotype Th2. The obtained results demonstrated that a PEEK-WC-HF membrane bioreactor is able to support the proliferation and functions of human peripheral lymphocytes isolated from the buffy coat of healthy individuals. Therefore, the lymphocyte HF membrane bioreactor can be used as a valuable tool to maintain viable and functional lymphocytes and as an in-vitro model for pharmacological and adoptive immunotherapy.

19.4 Membranes for Neuronal-Tissue Reconstruction

During the last few years neuronal cell behavior on a biomaterial such as membrane has become of great interest, since it offers the advantage of developing neuronal tissue that may be used for the in-vitro simulation of human brain functions. This could deﬁnitely provide further insights not only into the cell but also in developing therapies in neurodegenerative disorders such as Parkinsons or Alzheimers disease [51]. A biohybrid system using neurons could also represent a useful instrument for predictive drug testing or constitute a future model of a bioneuronal network device. For typical neuronal tissue-engineered constructs, the properties of both cell (morphology, viability functions) and material (physicochemical, morphological and transport properties) components are very important [52]. Current interest has been focused on attempts to ﬁnd new biomaterials and new cell

19.4 Membranes for Neuronal-Tissue Reconstruction

sources as well as novel designs of tissue-engineered neuronal devices to generate safer and more efﬁcacious restored neuronal tissue. Biomaterials that have been successfully employed in the manufacture of neuronal tissue include biodegradable materials such as polyglycolic acid, poly-L-lactic acid and poly (lactide-co-glycolide) and polymeric semipermeable membranes (polyacrylonitrile-polyvinylchloride, polyethylenecovinylalcohol, cellulose acetate) in ﬁber and ﬂat conﬁgurations [53–56]. In view of the widespread structural organization of all the brain regions, we focused our attention on the potential value of this biotechnological approach to a functionally key region such as the hippocampus. Indeed, the principal neurons of this brain region, that is, pyramidal cells, are actively involved in many hippocampaldependent neurophysiological functions, such as memory and learning. This makes them a valuable tool to investigate not only their synaptic plasticity properties, but also neurodegenerative events through the distribution and quantiﬁcation of microtubule-associated protein type 2. In a recent study, the reconstruction of membrane biohybrid systems, constituted of isolated cells and membranes, appears to represent a crucial step for the success of these systems [57]. Moreover, the optimization of transport, physicochemical and structural properties of the membrane as well as ﬂuid dynamics of cellular microenvironments tend to favor cell–membrane interactions and the functional maintenance of hippocampal cells. As a consequence, the feasibility of developing a hippocampal cell membrane biohybrid system capable of regenerating a neuronal network could prove to be an important approach for studying the behavior of neuronal populations in some of the most common neurodegenerative disease such as Alzheimers disease. A protocol of isolation and culture of hippocampal neurons has been optimized by using a hibernating rodent, the hamster Mesocricetus auratus as our animal model [57]. Preliminary results demonstrated the feasibility of culturing hippocampal neurons in a membrane biohybrid system. The substrate is of great importance for the survival and differentiation of neuronal cultures. As substrate we used a gas-permeable (CO2, O2, and H2O vapor) ﬂuorocarbon foil membrane (FC), modiﬁed with a coating of poly-L-lysine in order to improve the interactions with cells. In particular, it is interesting to note that after cells have adhered to the substrate a ﬂattening of the cells was observed and minor processes started to emerge from several sites along the circumference of the cells (Figure 19.6(a)). With the progress of their growth process, the tiny neuronal ﬁlaments begin to acquire the deﬁnite characteristics of dendrites and axons (Figure 19.6(b)) and subsequently the synaptic contacts in this rich neuronal network. The complexity of the neuronal network increased with time: dendrites emerging from the cell body became highly branched (Figure 19.6(c)). In our membrane biohybrid system, following the evaluation of some properties such as adhesion and development of dendrites and axons, the cytoskeletal features were considered by investigating the localization of bIII-tubulin (green) in the neuronal network (Figure 19.7). This cytoskeletal protein is present on the soma and in all neuronal processes. A complex axonal network was observed on the membrane (Figure 19.7) where the cells

j441

j 19 Membranes in Regenerative Medicine and Tissue Engineering

442

Figure 19.6 Micrographs of hippocampal neurons on FC membrane after (a) 4 hours, (b) 4 days (c) 16 days of culture. The arrows in a) indicate the emerging processes from the cell circumference; the arrows in b) indicate the (black) axon and the dendrites (white).

19.5 Concluding Remarks

Figure 19.7 Confocal laser micrographs of hippocampal neurons after 16 days of culture on modified FC membranes. The cells were immunolabeled for bIII-tubulin (green) and cell nuclei were labeled with DAPI (blue).

showed a differentiated pyramidal somata as well as forming signiﬁcantly longer neurites and more elaborated dendritic arbors. This well-deﬁned shape of the hippocampal neurons is important for in-vivo studies owing to their highly remarkable synaptic plasticity capacity, which proves to be essential for the explication of some key neurophysiological functions of hippocampus, such as memory and learning [58]. These results suggest that membrane system is able to favor the differentiation of neuronal hippocampal cells. The bIII-tubulin immunoreactivity displayed the unaltered cytoskeletal characteristics of cells after their differentiation and maturational process. These results encourage the development of a membrane-engineering system of hippocampal neurons that are able to remodel and regenerate neural tissue in a well-controlled microenvironment. This experimental system might be a valuable model to investigate complex neuronal networks existing between some major neurotransmitter systems such as the histaminergic and glutammatergic system and the behavior of neuronal populations in some of the most common neurodegenerative disease such as Alzheimers disease.

19.5 Concluding Remarks

This study reports on the potentiality of applying a membrane biohybrid system in the ﬁeld of tissue engineering and regenerative medicine evidencing the crucial points in the in-vitro reconstruction of the physiological tissue model. A number of issues need to be addressed: the morphological and physicochemical properties of the membrane, the optimal density of immobilized cells, the interaction of cells with the membrane, the differentiation of cells as well as the maintenance of viability and

j443

j 19 Membranes in Regenerative Medicine and Tissue Engineering

444

metabolic functions in vitro membrane constructs. Interesting results were obtained using membrane biohybrid systems for the in-vitro liver reconstruction, neuronal tissue, and for the in-vitro maintenance of the patients lymphocytes. The use of all these systems in tissue engineering can be exciting in helping to ﬁnd natures substitutes and in solving the pathogenesis of important human diseases or to select an optimal pharmaceutical treatment.

Acknowledgments

The authors acknowledge the Italian Ministry of University and Research (MIUR) for funding this research through the FIRB RBNE012B2K research project and the European Commission through the Livebiomat project STRP NMP3-CT-013653 in the FP6.

References 1 Hench, L.L. and Polak, J.M. (2002) Science, 295, 1014–1017. 2 De Bartolo, L., Jarosch-Von Schweder, G., Haverich, A. and Bader, A. (2000) Biotechnology Progress, 16, 102–108. 3 De Bartolo, L., Morelli, S., Bader, A. and Drioli, E. (2002) Biomaterials, 23, 2485–2497. 4 Unger, R.E., Huang, Q., Peters, K., Protzer, D., Paul, D. and Kirkpatrick, C.J. (2005) Biomaterials, 26, 1877–1884. 5 Gentile, F.T., Doherty, E.J., Rein, D.H., Shoichet, M.S. and Winn, S.R. (1995) Reactive Polymers, 25, 207–227. 6 Li, R.H. (1998) Advanced Drug Delivery Reviews, 33, 87–109. 7 Quirk, R.A., Chan, W.C., Davies, M.C., Tendler, S.J.B. and Shakesheff, K.M. (2001) Biomaterials, 22, 865–872. 8 Hubbell, J.A., Massia, S.P., Desai, N.P. and Drumheller, P.D. (1991) Biotechnology (Reading, Mass), 9, 568–574. 9 Hunter, D.D., Cashman, N., MorrisValero, R., Bulock, J.W., Adams, S.P. and Sanes, J.R. (1991) The Journal of Neuroscience, 11, 3960–3971. 10 Patel, N., Padera, R., Sanders, H.W., Cannizzaro, M., Davies, C., Langer, R., Roberts, J., Tendler, J.B., Williams, M. and

11 12

13

14

15

16

17

18

Shakesheff, M. (1998) The FASEB Journal, 12, 1447–1454. Michalopoulos, G.K. and DeFrances, M.C. (1997) Science, 276, 60–66. Berry, M.N. and Edwards, A.M. (2000) The Hepatocyte Review, Kluwer Academic Publishers, Dordrecht, pp. 365–585. Dunn, J.C., Tompkins, R.G. and Yarmush, M.L. (1991) Biotechnology Progress, 7, 237–244. De Bartolo, L., Morelli, S., Bader, A. and Drioli, E. (2001) Journal of Material Science: Materials in Medicine, 12, 959–963. De Bartolo, L., Morelli, S., Rende, M., Gordano, A. and Drioli, E. (2004) Biomaterials, 25, 3621–3629. De Bartolo, L., Morelli, S., Gallo, M.C., Campana, C., Statti, G., Rende, M., Salerno, S. and Drioli, E. (2005) Biomaterials, 26, 6625–6634. De Bartolo, L., Morelli, S., Lopez, L.C., Giorno, L., Campana, C., Salerno, S., Rende, M., Favia, P., Detomaso, L., Gristina, R., dAgostino, R. and Drioli, E. (2005) Biomaterials, 26, 4432–4441. De Bartolo, L., Morelli, S., Giorno, L., Campana, C., Rende, M., Salerno, S., Maida, S. and Drioli, E. (2006) Journal of Membrane Science, 278 (1–2), 133–143.

References 19 De Bartolo, L., Salerno, S., Giorno, L., Morelli, S., Barbieri, G., Curcio, E., Rende, M. and Drioli, E. (2006) Catalysis Today, 118, 172–180. 20 De Bartolo, L., Morelli, S., Rende, M., Campana, C., Salerno, S., Quintiero, N. and Drioli, E. (2007) Macromolecular Bioscience, 7, 671–680. 21 Memoli, B., De Bartolo, L., Favia, P., Morelli, S., Lopez, L.C., Procino, A., Barbieri, G., Curcio, E., Giorno, L., Esposito, P., Cozzolino, M., Brancaccio, D., Andreucci, V.E., dAgostino, R. and Drioli, E. (2007) Biomaterials, 28, 4836–4844. 22 De Bartolo, L., Morelli, S., Rende, M., Salerno, S., Giorno, L., Lopez, L.C., Favia, P., dAgostino, R. and Drioli, E. (2006) Journal of Nanoscience and Nanotechnology, 6, 2344–2353. 23 De Bartolo, L., Morelli, S., Piscioneri, A., Lopez, L.C., Favia, P., dAgostino, R. and Drioli, E. (2007) Biomolecular Engineering, 24, 23–26. 24 Morelli, S., Salerno, S., Rende, M., Lopez, L.C., Favia, P., Procino, A., Memoli, B., Andreucci, V.E., dAgostino, R., Drioli, E. and De Bartolo, L., (2007) Journal of Membrane Science. 302, 27–35. 25 Ying, L., Yin, C., Zhuo, R.X., Leong, K.W., Mao, H.Q., Kang, E.T. and Neoh, K.G. (2003) Biomacromolecules, 4, 157–164. 26 Lopina, S.T., Wu, G., Merrill, E.W. and Grifﬁth, C.L. (1996) Biomaterials, 17, 559–567. 27 Rocha, F.G. and Whang, E.E. (2004) Journal of Surgical Research, 120, 320–325. 28 Sodian, R., Hoerstrup, S.P., Sperling, J.S., Daebritz, S.H., Martin, D.P., Shoen, F.J., Vacanti, J.P. and Mayer, J.E. (2000) The Annals of Thoracic Surgery, 70, 140–144. 29 Shinoka, T., Imai, Y. and Ikada, Y. (2001) The New England Journal of Medicine, 3444, 532–540. 30 Fansa, H., Dodic, T., Wolf, G., Schneider, W. and Keilhoff, G. (2003) Microsurgery, 23, 72–78. 31 Parenteau, N.L., Nolte, C.M., Bilbo, P., Rosenberg, M., Wilkins, L.M., Johnson, E.W., Watson, S., Mason, V.S. and Bell, E.J.,

32

33

34 35

36

37

38

39

40

41

42

43 44

45

46

(1991) Journal of Cellular Biochemistry, 45, 245–253. Vacanti, C.A. and Vacanti, J.P. (1994) Otolaryngologic Clinics of North America, 27, 263–269. Tsai, R.J., Li, L.M. and Che, J.K. (2000) The New England Journal of Medicine, 343, 86–95. Jauregui, H.O. and Gann, K.L. (2002) Tissue Engineering, 8, 725–732. Papas, K.K., Long, R.C., Sambanis, A. and Constantinidis, I. (1999) Biotechnology and Bioengineering, 66, 219–226. Dionne, K.E., Cain, B.M., Li, R.H., Bell, W.J., Doherty, E.J., Rein, D.H., Lysaght, M.J. and Gentile, F.T. (1996) Biomaterials, 17, 257–266. Curcio, E., De Bartolo, L., Barbieri, G., Rende, M., Giorno, L., Morelli, S. and Drioli, E. (2005) Journal of Biotechnology, 117, 309–318. Stenvinkel, P., Ketteler, M., Johnson, R.J., Lindholm, B., Pecoits-Filho, R., Heimburger, O., Cederholm, T., Girudt, M. and Riella, M. (2005) Kidney International, 67, 1216–1233. Bremers, A.J.A. and Parmiani, G. (2000) Critical Reviews in Oncology/Hematology, 34, 1–25. Trickett, A.E., Kwan, Y.L., Cameron, B. and Dwyer, J.M. (2002) Journal of Immunological Methods, 262, 71–83. Krovat, B.C., Tracy, J.H. and Omiecinski, C.J. (2000) Toxicological Sciences, 55, 352–360. Tabatabetei, A.R., Thies, R.L., Farrell, K. and Abbott, F.S. (1997) Fundamental and Applied Toxicology, 37, 181–189. Muehlbauer, P.A. and Schuler, M.J. (2003) Mutation Research, 537, 117–130. Garland, R.J., Kaneria, S.S., Hancock, J.P., Steward, C.G. and Rowbottom, A.W. (1999) Journal of Immunological Methods, 227, 53–63. Woodside, K.J., Hu, M., Gugliuzza, K.K., Hunter, G.C. and Daller, J.A., (2005) Transplantation Proceedings, 37, 1949–1952. Martin, Y. and Vermette, P. (2005) Biomaterials, 26, 7481–7503.

j445

j 19 Membranes in Regenerative Medicine and Tissue Engineering

446

47 Lamers, C.H.J., Gartama, J.W., LuiderVrieling, B., Bolhuis, R.L.H. and Bast, E.J.E.G. (1999) Journal of Immunotherapy, 22, 4, 299–307. 48 Knazek, R.A., Wu, Y.W., Aebersold, P.M. and Rosenberg, S.A. (1990) Journal of Immunological Methods, 127, 29–37. 49 Cadwell, J.J.S. (2004) American Biotechnology Laboratory, 1–5. 50 Bohnenkamp, H., Hilbert, U. and Noll, T. (2002) Cytotechnology, 38, 135–145. 51 Caffesse, R.G., Najleti, C.E., Morrisson, E.C. and Sanchez, R. (1994) Journal of Periodontology, 65, 583–590. 52 Zhang, N., Yan, H. and Wen, X. (2005) Brain Research Reviews, 49, 48–56. 53 Young, T.H., Lin, C.W., Cheng, L.P. and Hsieh, C.C. (2001) Biomaterials, 22, 1771–1779.

54 Gautier, S.E., Oudega, M., Fragoso, M., Chapon, P., Plant, G.W., Bunge, M.B. and Parel, J.M. (1998) Journal of Biomedical Materials Research, 42, 642–648. 55 Patist, C.M., Mulder, M.B., Gautier, S.E., Maquet, V., Jerome, R. and Oudega, M. (2004) Biomaterials, 25, 1569–1576. 56 Woerly, S., Plant, G.W. and Harvey, A.R. (1996) Biomaterials, 17, 301–313. 57 De Bartolo, L., Rende, M., Giusi, G., Morelli, S., Piscioneri, A., Canonaco, M. and Drioli, E. (2007) in Evolutionary Molecular Strategies and Plasticity (eds M. Canonaco and R.M. Facciolo), Research Signpost, Kerala, India, pp. 379–396. 58 Lo, D.C. (1995) Neuron, 15 (5), 979–981.

Part Three Membrane Contactors This Part focuses on fundamentals and applications of membrane contactors and membrane emulsiﬁcation. Special attention is given to the industrial application of membrane contactors. Among other new unit operations involving membranes, membrane contactors are expected to play a decisive role in the development of clean, safe and sustainable processes. The key concept is to use a solid, microporous, polymeric matrix in order to create an interface for mass transfer and/or reaction between two phases: large exchange area and independent ﬂuid dynamics allow an easily controlled operation. In general, these membrane systems use low-cost hollow ﬁbers, and provide a high interfacial area signiﬁcantly greater than most traditional absorbers between two phases to achieve high overall rates of mass transfer. In addition, whereas the design of the conventional devices is restricted by limitations in the relative ﬂows of the ﬂuid streams, membrane contactors give an active area, which is independent of the liquid ﬂuid dynamics. Membrane crystallizers, membrane emulsiﬁers, membrane strippers and scrubbers, membrane distillation systems, membrane extractors, etc., can be devised and integrated in the production lines together with the other existing membranes operations for advanced molecular separation and chemical transformations, overcoming existing limits of the more traditional membrane processes, such as osmotic effect in reverse osmosis. In these systems, the interface between two phases is located at the highthroughput membrane porous matrix level. Physicochemical, structural and geometrical properties of porous meso- and microporous membranes are exploited to facilitate mass transfer between two contacting immiscible phases, e.g., gas–liquid, vapor–liquid, liquid–liquid, liquid–supercritical ﬂuid, etc., without dispersing one phase in the other (except for membrane emulsiﬁcation, where two phases are contacted and then dispersed drop by drop one into another under precise controlled conditions). Separation depends primarily on phase equilibrium. Membrane-based absorbers and strippers, extractors and back extractors, supported gas membranebased processes and osmotic distillation are examples of such processes that have already been in some cases commercialized. Membrane distillation, membrane

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

crystallization, supported liquid membranes, etc., are examples of such processes that are not yet commercialized. Indeed, they are of great interest due to their inherent capabilities. The integration of membrane contactors with other membrane operations, including membrane reactors, may lead to redesign of production lines based on intensiﬁed integrated membrane processes.

j449

20 Basics in Membrane Contactors Alessandra Criscuoli

20.1 Introduction

This chapter aims at providing an overview on membrane contactors, starting from their deﬁnition and main properties, going through the mass transport, and ﬁnally describing their main applications. In particular, the role of membrane contactors in reaching the goals of the process-intensiﬁcation strategy is discussed in terms of new developed metrics. The objective is to give some basic information on these new systems of mass transfer and their potentialities, as well as on the main drawbacks related to their further implementation at the industrial level and on the research efforts to be made for overcoming these limitations and to extend their ﬁelds of application.

20.2 Definition of Membrane Contactors

A membrane is usually seen as a selective barrier that is able to be permeated by some species present into a feed while rejecting the others. This concept is the basis of all traditional membrane operations, such as microﬁltration, ultraﬁltration, nanoﬁltration, reverse osmosis, pervaporation, gas separation. On the contrary, membrane contactors do not allow the achievement of a separation of species thanks to the selectivity of the membrane, and they use microporous membranes only as a mean for keeping in contact two phases. The interface is established at the pore mouths and the transport of species from/to a phase occurs by simple diffusion through the membrane pores. In order to work with a constant interfacial area, it is important to carefully control the operating pressures of the two phases. Usually, the phase that does not penetrate into the pores must be kept at higher pressure than the other phase (Figure 20.1a and b). When the membrane is hydrophobic, polar phases can not go into the pores, whereas, if it is hydrophilic, the nonpolar/gas phase remains blocked at the pores entrance [1, 2].

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 20 Basics in Membrane Contactors

450

Figure 20.1 Contact between phase 1 and phase 2 through a membrane contactor. (a) phase 1 in the membrane; (b) phase 2 in the membrane.

Membrane contactors can be seen as alternative systems for carrying out gas–liquid operations (such as stripping and scrubbing) or liquid–liquid extractions. Furthermore, by applying vacuum at one membrane side, liquid–vacuum operations can be performed. Another interesting process that can be carried out is distillation (up to crystallization). In particular, depending on the way the driving force is obtained, we can distinguish between membrane and osmotic distillation. In direct contact membrane distillation (DCMD), which has been the most investigated conﬁguration, a difference of temperature is imposed between the two streams, whereas in osmotic distillation, the driving force is achieved by using a hypertonic solution (often, a highly concentrated CaCl2 solution) as strip phase. Finally, supported liquid membranes represent a particular type of membrane contactors where, usually, an organic phase ﬁlls the micropores and the transport occurs from an aqueous feed to an aqueous strip. Figure 20.2 summarizes the different types of membrane contactors described. Membrane-contactor efﬁciency depends on membrane properties, membrane module, operating conditions. Hydrophobic membranes are often used and, among the hydrophobic polymeric materials, we can mention polypropylene, PVDF, PTFE, and perﬂuoropolymers (e.g., hyﬂon). For a good performance of the system it is important that the membrane keeps its hydrophobic character, especially in long-term applications. Therefore, several studies aimed at improving the hydrophobicity values, as well as increasing the membrane stability with time, have been made [3–6]. The chemical stability of the membrane is also crucial for successful applications with organic solvent or absorbents (such as amines). Membranes with big pore sizes, high porosities and low thicknesses lead to a high transfer of the species, but big pore sizes also mean lower values of the breakthrough pressures (pressures at which the membrane is wetted by the liquid, losing its hydrophobic character), as reported in Laplaces equation (for gas–liquid operations): Dp ¼ ð2scosqÞ=r

ð20:1Þ

with s, surface tension; q contact angle; r, pore radius. Moreover, in direct contact membrane distillation, a minimum value of thickness is required to keep the difference of temperature across the membrane. Generally, in membrane distillation materials with low thermal conductivity are also required to reduce the heat loss through the membrane-self. Pore-size distribution also plays an

20.2 Definition of Membrane Contactors

Figure 20.2 Different types of membrane contactors. (a) stripper/ scrubber, driving force: difference of concentration; (b) liquid–liquid extractors, driving force: difference of concentration; (c) removal of volatiles/gases from liquids, driving force: difference of partial pressures; (d) direct contact membrane distillation, driving force: difference of partial pressures; (e) supported liquid membranes, driving force: difference of concentrations.

important role for obtaining a uniform transport along the membrane and avoiding coalescence phenomena. Table 20.1 summarizes the main effects of some membrane properties on the performance of membrane contactors. Referring to the membrane module design, it has a big inﬂuence on the membrane-contactor efﬁciency, because it affects the pressure drops of the streams (and, thus, the operating pressures and ﬂowrates), and their ﬂuidodynamic (which means the mass and heat transport resistances of the phases). Furthermore, for hollow-ﬁber modules it is essential to ensure a uniform packing, in order to have

j451

j 20 Basics in Membrane Contactors

452

Table 20.1 Main effects of some membrane properties on the

performance of membrane contactorsa. Membrane property

Effect

Big pore size High porosity Low thickness

Higher ﬂux (þ) and lower breakthrough pressure value () Higher ﬂux (þ) and higher coalescence phenomena () Higher ﬂux (þ) and lower difference of temperature across the membrane (), for DCMD Higher ﬂux (þ)

Low tortuosity

Symbols ( þ ) and () indicates a positive and a negative effect, respectively.

a

an uniform ﬂow pattern at the shell side, avoiding any phenomena of channeling, bypassing, presence of stagnant areas, which can sensibly reduce the overall performance. For these reasons there have been several studies for developing modules for membrane-contactor applications [7–13]. The right choice of operating conditions is also at the basis of a good performance of membrane contactors. Higher ﬂowrates lead to a reduction of the mass and heat transport resistances in the phases and of the presence of stagnant zones inside the module. However, they have to be carefully deﬁned, in order to avoid the stream pressures reaching the breakthrough values. For the same reason, the pressures of the streams to be processed must be controlled, and eventually properly varied, before sending them to the membrane-contactor unit. Streams with high viscosity or containing particles of big size should be preferably sent to the shell side of the module, and the ﬂuid with higher afﬁnity for the species to be transferred should ﬁll the membrane pores, in order to reduce the membrane mass-transfer resistance. When compared to conventional systems (such as strippers, scrubbers, distillation columns, packed towers, bubble columns, evaporators, etc.), membrane contactors present several advantages, as reported in Figure 20.3. However, some drawbacks have also to be taken into account, as shown in Figure 20.4.

20.3 Mass Transport

When a species is transferred from a phase to another phase by means of a membrane contactor, the mass-transport resistances involved are those offered by the two phases and that of the membrane (see Figure 20.5). The overall mass-transfer coefﬁcient will, therefore, depend on the mass-transfer coefﬁcient of the two phases and of the membrane. If we consider a gas–liquid transfer for the species i in a hollow-ﬁber module with the liquid phase in the shell side and the gas phase in the lumen side of hydrophobic membranes, the interface is established at the outer diameter of the ﬁbers and the overall mass-transfer coefﬁcient can be calculated by [1]: 1=ðKl do Þ ¼ 1=ðkils do Þ þ 1=ðkim Hi dlm Þ þ 1=ðkigt Hi di Þ

ð20:2Þ

20.3 Mass Transport

Figure 20.3 Main advantages of membrane contactors.

1=ðKg do Þ ¼ Hi =ðkils do Þ þ 1=ðkim dlm Þ þ 1=ðkigt di Þ

ð20:3Þ

where Kl, liquid overall mass-transfer coefﬁcient; Kg, gas overall mass-transfer coefﬁcient; kils, mass-transfer coefﬁcient for the species i in the liquid at the shell side; kigt, mass-transfer coefﬁcient for the species i in the gas at the tube side; kim, mass-transfer coefﬁcient for the species i in the membrane; di, inner diameter of the ﬁber; do, outer diameter of the ﬁber; dlm, logarithmic mean of the hydrophobic membrane diameters; Hi, Henrys coefﬁcient. The membrane mass-transfer coefﬁcient kim is usually based on Knudsen ﬂows and can be derived by: kim ¼ Dki e=td

ð20:4Þ

with Dki , Knudsen diffusion coefﬁcient for the species i through the membrane; e, membrane porosity; t, membrane tortuosity; d, membrane thickness. The mass-transfer coefﬁcient at the tube side is usually well obtained by the Leveque equation: Sh ¼ 1:62 ðd2 v=ðLDÞÞ0:33

ð20:5Þ

j453

j 20 Basics in Membrane Contactors

454

Figure 20.4 Main disadvantages of membrane contactors.

Figure 20.5 Mass-transport resistances involved in a membrane contactor.

20.4 Applications

where Sh, Sherwood number; v, ﬂuid velocity; d, ﬁber diameter; L, length of the ﬁber; D, diffusion coefﬁcient of the species i into the ﬂuid. On the contrary, no general expression is available for calculating the mass-transfer coefﬁcient at the shell side. In the literature, in fact, different equations are proposed, depending on the type of module and on the type of ﬂow (parallel or crossﬂow). Probably, this is due to the fact that the ﬂuidodynamics of the stream sent outside the ﬁbers is strongly affected by the phenomena of channeling or bypassing and it is not well deﬁned as for the stream, which is sent into the ﬁbers. Hereinafter some of the different expressions proposed are reported. Parallel ﬂow Sh ¼ b de =L ð1fÞ Re0:60 Sc0:33

½14

ð20:6Þ

0 < Re < 500; loosely packed ﬁbers; b is 5.8 for hydrophobic and 6.1 for hydrophilic membranes Sh ¼ ð0:31f2 0:34f þ 0:10ÞRe0:90 Sc0:33

½10

ð20:7Þ

Re ¼ 100; loosely closely packed ﬁbers Sh ¼ ð0:530:58fÞ Re0:53 Sc0:33

½15

ð20:8Þ

21 < Re < 324; medium-closely packed ﬁbers with Re, Reynolds number; Sc, Schmidt number; f, ﬁber packing; de, equivalent diameter. Crossﬂow Sh ¼ 0:15 Re0:80 Sc0:33 Sh ¼ 0:12 Re Sc0:33

Re > 2:5

Re < 2:5

Sh ¼ 1:38 Re0:34 Sc0:33

½13

½13

ð20:9Þ ð20:10Þ

½16

ð20:11Þ

1 < Re < 25; closely packed ﬁbers Sh ¼ 0:90 Re0:40 Sc0:33

½16

ð20:12Þ

1 < Re < 25; loosely packed ﬁbers Sh ¼ fRe1:2 Sc0:33

0:03 < Re < 0:3

½17

ð20:13Þ

20.4 Applications

The variety of unit operations that can be performed by membrane contactors has led during last years to many research studies in which membrane contactors are applied to different ﬁelds.

j455

j 20 Basics in Membrane Contactors

456

In particular, liquid–liquid extractions, wastewater treatments, gas absorption and stripping, membrane, and osmotic distillation, are the processes more studied. For example, the VOCs removal, the extraction of aroma compounds and metal ions, the concentration of aqueous solutions, the acid-gases removal, the bubble-free oxygenation/ozonation, have been successfully carried out by using membrane contactors [1, 2]. The design of the ﬁrst commercial modules has allowed the commercial application of membrane contactors for some speciﬁc operations. This is the case of the Membrana-Charlotte Company (USA) that developed the LiquiCel modules, equipped with polypropylene hollow ﬁbers, for the water deoxygenation for the semiconductor industry. LiquiCel modules have been also applied to the bubble-free carbonation of Pepsi, in the bottling plant of West Virginia [18], and to the concentrations of fruit and vegetable juices in an osmotic distillation pilot plant at Melbourne [19]. Other commercial applications of LiquiCel are the dissolved-gases removal from water, the decarbonation and nitrogenation in breweries, and the ammonia removal from wastewater [20]. By using the rectangular transverse-ﬂow module developed by TNO (The Netherlands), an industrial membrane gas absorption unit for ammonia recovery has been installed in The Netherlands [21]. Also, a unit for CO2 removal has been tested [22]. Flat-membrane contactors have been speciﬁcally designed and commercialized by GVS SpA (Italy) for air dehumidiﬁcation processes [23]. As previously reported, membrane contactors present interesting advantages with respect to traditional units. Moreover, they well respond to the main targets of the process intensiﬁcation, such as to develop systems of production with lower equipment-size/production-capacity ratio, lower energy consumption, lower waste production, higher efﬁciency. In order to better identify the potentialities of membrane contactors in this logic, they have been recently compared to traditional devices for the sparkling-water production in terms of new deﬁned indexes [24]. In particular, the comparison has been made at parity of plant capacity and quality of ﬁnal product. The metrics used for the comparison between membranes and traditional units are: (a) the productivity/size ratio (PS); (b) the productivity/weight ratio (PW); (c) the ﬂexibility, as the ability to handle the changes in the operating conditions (ﬂexibility1); (d) the ﬂexibility, as the ability to cover different applications (ﬂexibility2); (e) the modularity; (f) the mass intensity ratio. In Table 20.2 the results of the study are shown. From this, it results that when the process is carried out by membrane contactors there is a gain in terms of ﬂexibility and modularity of the plant, and lower CO2 consumptions and size and weight are achieved. Another ﬁeld where membrane contactors are studied in the logic of the process intensiﬁcation is seawater and brackish-water desalination. The scarcity of

20.4 Applications Table 20.2 Comparison between traditional and proposed system

for sparkling-water production [24]. Traditional system (deareator þ saturator)

Membrane contactor

PS PW Flexibility1 Flexibility2 Modularity Mass intensity ratio

6.46 2.33 1.39 3 <1 0.85

fresh-water resources is today becoming a big issue worldwide and there are the several studies in progress for improving the existing desalination plants in terms of higher fresh-water recovery, lower use of chemicals, lower brine production, higher quality of the produced water [25–27]. In a desalination plant, membrane contactors can be introduced for different purposes: (a) for controlling the oxygen/carbon dioxide content of the water; (b) for disinfection/oxidation of the water; (c) for reducing the boron content into the water, working on the reverse osmosis permeate as liquid–liquid device; (d) for improving the fresh-water recovery factor of reverse osmosis, working as a membrane distillation unit; (e) for producing crystals from brine, working as a membrane crystallizer. Referring to point a, in desalination the content of oxygen and carbon dioxide in the water affects the material life of the plant (because of corrosion problems), as well as the pH and the conductivity of water. Usually, these gases are removed by stripping in a packed column and the ﬁnal water pH is adjusted by means of caustic soda. By using membrane contactors, there is no need for chemicals, with a consequent reduction of the environmental impact. Membrane contactors can be effectively used also for disinfection purposes (e.g., water ozonation) [28] or for the oxidation of species present into water, for example, arsenic. Although the content of arsenic in seawater is today within the accepted limits, it is foreseen that in the future its concentration could increase, due to the increase of pollution of rivers and groundwaters. Usually, arsenic is contained in water as As(III) and As(V) forms, in different amounts. All arsenic-removal technologies have a better performance when arsenic is present in the pentavalent

j457

j 20 Basics in Membrane Contactors

458

form, therefore, membrane contactors can be used as an alternative system for oxidizing As(III) to As(V), by simply using air. From preliminary results obtained at the lab-scale, membrane contactors seem to be quite effective for achieving a complete oxidation [29]. The control of the boron content into water is another big issue because the existing reverse-osmosis membranes are not able to readily reject boron at natural pHs, thus several stages in series, operating at different pHs, coupled also to selective boron-exchange resins, are required for achieving the desired boron content [30]. Recently, an alternative desalination ﬂowsheet where the permeate coming from the reverse-osmosis unit (usually containing 1 ppm of boron) is treated in a membrane-contactor device has been proposed and designed [31]. The boron removal occurs by simple diffusion across the hydrophilic microporous membranes from the feed stream to another water stream (stripping phase) that is recycled to the membrane contactor unit, after its puriﬁcation in a BWRO device (see Figure 20.6). From a comparison made in terms of the new metrics, between the proposed scheme and that traditionally employed, at parity of fresh-water production and ﬁnal boron content, it results that the reverse-osmosis-membrane contactor integrated system is advantageous in terms of size, energy consumption, use of chemicals, ﬂexibility and modularity [31]. The fact that the performance of membrane distillation is not strongly affected by the concentration of the solution to be treated, allows to work with the membrane distillation unit on the reverse-osmosis concentrate, in order to further increase the

Figure 20.6 An integrated reverse-osmosis-membrane contactor system for boron removal (From Ref. [31].)

20.4 Applications

Figure 20.7 NaCl crystals obtained in a membrane crystallizer (magnification: 10) (From Ref. [26].)

fresh-water recovery and to reduce the brine to be disposed. In particular, the membrane distillation can be used for obtaining supersaturated brines, to be further processed in membrane crystallizers for producing valuable crystals. In this way, the problem of the brine disposal is completely avoided, with a consequent reduction of the environmental impact. As example, Figures 20.7 and 20.8 show pictures of NaCl and MgSO47H2O crystals, respectively, obtained in a membrane crystallizer [26]. The membrane distillation units and membrane crystallizers will be also energy efﬁcient if sources of energy such as solar are considered for providing the heat required for the water evaporation.

Figure 20.8 MgSO47H2O crystals obtained in a membrane crystallizer (magnification: 20) (From Ref. [26].)

j459

j 20 Basics in Membrane Contactors

460

20.5 Concluding Remarks

The introduction of membrane contactors in industrial cycles might represent an interesting way to realize the rationalization of chemical productions in the logic of the process intensiﬁcation. Membrane contactors are, in fact, highly efﬁcient systems for carrying out the mass transfer between phases and achieving high removals. They also present lower size than conventional apparatus. Commercial applications are already present (e.g., the electronics industry or bubble-free carbonation lines), however, some critical points must be still overcome and several are the research efforts needed for their further implementation at industrial level, as summarized below: (a) the membrane introduces another resistance to the mass transfer with respect to traditional systems: thin membranes have to be used; (b) shell-side bypassing results in a loss of efﬁciency, especially for big contactors: design improvements are required; (c) high hydrophobicity and methods for restoring it are required for successful applications; (d) in order to reduce the heat loss through the membrane, materials with low thermal conductivity are required in membrane distillation operations; (e) the operating pressures are limited by the breakthrough pressure value: membranes able to withstand high pressures will extend the range of membranecontactor applications; (f) more general expressions to calculate the shell-side mass-transfer coefﬁcients are required in order to facilitate the scale-up.

References 1 Drioli, E., Criscuoli, A. and Curcio, E. (2006) Membrane Contactors: Fundamentals, Applications and Potentialities, Elsevier, Amsterdam, 0-444-52203-4. 2 Gabelman, A. and Hwang, S.T. (1999) Journal of Membrane Science, 159, 61–106. 3 Li, K., Wang, D., Koe, C.C. and Teo, W.K. (1998) Chemical Engineering Science, 53 (6), 1111–1119. 4 Bhaumik, D., Majumdar, S. and Sirkar, K.K. (2000) Journal of Membrane Science, 167, 107–122. 5 Majumdar, S., Bhaumik, D. and Sirkar, K.K. (2003) Journal of Membrane Science, 214, 323–330. 6 Gugliuzza, A. and Drioli, E. (2007) Journal of Membrane Science, 300, 51–62.

7 Ghogomu, J.N., Guigui, C., Rouch, J.C., Clifton, M.J. and Aptel, P. (2001) Journal of Membrane Science, 81, 71–80. 8 Wang, K.L. and Cussler, E.L. (1993) Journal of Membrane Science, 85, 265–278. 9 Seibert, A.F. and Fair, J.R. (1997) Separation Science and Technology, 32, 1, 573–583. 10 Wu, J. and Chen, V. (2000) Journal of Membrane Science, 172, 59–74. 11 Lemanski, J. and Lipscomb, G.G. (2001) Journal of Membrane Science, 195, 215–228. 12 Lemanski, J., Liu, B. and Lipscomb, G.G. (1999) Journal of Membrane Science, 153, 33–43. 13 Wickramasinghe, S.R., Semmens, M.J. and Cussler, E.L. (1992) Journal of Membrane Science, 69, 235–250.

References 14 Prasad, R. and Sirkar, K.K. (1998) AICHE Journal, 34 (2), 177–188. 15 Costello, M.J., Fane, A.G., Hogan, P.A. and Schoﬁeld, R.W. (1993) Journal of Membrane Science, 80, 1–11. 16 Yang, M.-C. and Cussler, E.L. (1986) AIChE Journal, 32, 1910–1915. 17 Criscuoli, A., Drioli, E. and Moretti, U. (2003) Annals of the New York Academy of Sciences, 984, 1–16. 18 Peterson, P.A., Schneider, J. and Sengupta, A. (1998) Proceedings of the Workshop on Membrane Distillation, Osmotic Distillation and Membrane Contactors, Cetraro (CS) Italy, July 2–4. 19 Hogan, P.A., Canning, R.P., Peterson, P.A., Johnson, R.A. and Michaels, A.S. (1998) Chemical Engineering Progress, 94, 49–61. 20 www.liqui-cel.com. 21 Klaassen, R., Feron, P.H.M. and Jansen, A.E. (2005) Chemical Engineering Research & Design, 83 (A3), 234–246. 22 Feron, P. and Jansen, A.E. (2002) Separation and Puriﬁcation Technology, 27, 231–242. 23 Gaeta, S.N. (2003) Proceedings of the 1st Workshop Italy–Russia, Cetraro (CS) Italy, September 17–20.

24 Criscuoli, A. and Drioli, E. (2007) Industrial & Engineering Chemistry Research, 46, 2268–2271. 25 Drioli, E., Criscuoli, A. and Curcio, E. (2002) Desalination, 147, 77–81. 26 Drioli, E., Curcio, E., Criscuoli, A. and Di Proﬁo, G. (2004) Journal of Membrane Science, 239, 27–38. 27 Drioli, E., Curcio, E., Di Proﬁo, G., Macedonio, F. and Criscuoli, A. (2006) Trans IChemE, Part A, Chem Eng Res and Des, 84 (A3), 209–220. 28 Cissel, J., Gramer, M., Shanbhag, P.V. and Nemser, S.M. (March 5–9, 2000) Proceedings of the AIChE Spring National Meeting Atlanta, pp. 56–62. 29 Galizia, A.(May 2007) Master thesis on Treatment of water containing arsenic by means of membrane operations, University of Calabria (Italy). 30 Redondo, J., Busch, M. and De Witte, J.-P. (2003) Desalination, 156, 229–238. 31 Criscuoli, A., Rossi, E., Cofone, F. and Drioli, E., (2009) accepted by Clean Technologies and Environmental Policy.

j461

j463

21 Membrane Emulsification: Principles and Applications Lidietta Giorno, Giorgio De Luca, Alberto Figoli, Emma Piacentini, and Enrico Drioli

21.1 Introduction

Emulsions and suspensions are colloidal dispersions of two or more immiscible phases in which one phase (disperse or internal phase) is dispersed as droplets or particles into another phase (continuous or dispersant phase). Therefore, various types of colloidal systems can be obtained. For example, oil/water and water/oil single emulsions can be prepared, as well as so-called multiple emulsions, which involve the preliminary emulsiﬁcation of two phases (e.g., w/o or o/w), followed by secondary emulsiﬁcation into a third phase leading to a three-phase mixture, such as w/o/w or o/w/o. Suspensions where a solid phase is dispersed into a liquid phase can also be obtained. In this case, solid particles can be (i) microspheres, for example, spherical particles composed of various natural and synthetic materials with diameters in the micrometer range: solid lipid microspheres, albumin microspheres, polymer microspheres; and (ii) capsules, for example, small, coated particles loaded with a solid, a liquid, a solid–liquid dispersion or solid–gas dispersion. Aerosols, where the internal phase is constituted by a solid or a liquid phase dispersed in air as a continuous phase, represent another type of colloidal system. In emulsions and suspensions, disperse phase dimensions may vary from the molecular state to the coarse (visible) dispersion. They are commonly encountered in various productions. The average droplet/microcapsules size distribution is a key feature since they determine emulsions/suspensions properties for the intended uses and stability. For large-scale emulsion production, the most commonly employed methods are based on techniques aiming at establishing a turbulent regime in the ﬂuid mixtures. These turbulent ﬂows cannot be controlled or generated uniformly. The consequences are that the control of the droplet sizes is difﬁcult and wide size distributions are commonly obtained, therefore the energy is used inefﬁciently in these technologies. In addition, the process scale-up is extremely difﬁcult. The use of the ultrasonic bath yields better results with respects to the

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 21 Membrane Emulsification: Principles and Applications

464

mentioned procedures, however, the control of the droplet dimension is still not optimal. For these reasons, recently much attention has been put in alternative emulsiﬁcation processes, such as the membrane emulsiﬁcation (ME). Membrane emulsiﬁcation is an appropriate technology for production of single and multiple emulsions and suspension. It was proposed for the ﬁrst time at the 1988 Autumn Conference of the Society of Chemical Engineering, Japan. Since then, the method has continued to attract attention in particular in Japan, but also in Europe [1–10]. In the early 1990s, Nakashima et al. [2] introduced membrane technology in emulsions preparation by a direct emulsiﬁcation method, whereas, in the late 1990s, Suzuki et al. used premix membrane emulsiﬁcation to obtain production rates higher than other membrane emulsiﬁcation methods [11]. The fast progress in microengineering and semiconductor technology led at the development of microchannels, that Nakajima et al. applied in emulsiﬁcation technology [12]. The distinguishing feature of membrane emulsiﬁcation technique is that droplet size is controlled primarily by the choice of the membrane, its microchannel structure and few process parameters, which can be used to tune droplets and emulsion properties. Comparing to the conventional emulsiﬁcation processes, the membrane emulsiﬁcation permits a better control of droplet-size distribution to be obtained, low energy, and materials consumption, modular and easy scale-up. Nevertheless, productivity (m3/day) is much lower, and therefore the challenge in the future is the development of new membranes and modules to keep the known advantages and maximize productivity. Considerable progress has been achieved in understanding the technology from the experimental point of view, with the establishment of many empirical correlations. On the other hand, their theoretical interpretation by means of reliable models is not accordingly advanced. The ﬁrst model devoted to membrane emulsiﬁcation, based on a torque balance, was proposed in 1998 by Peng and Williams [13], that is, ten years later the ﬁrst experimental work was published, and still nowadays, a theoretical study aiming at a speciﬁc description of the premix membrane emulsiﬁcation process is not available. The nonsynergistic progress of the theoretical understanding with the experimental achievements, did not refrain the technology application at the productive scale. In particular, membrane emulsiﬁcation was successfully applied for preparation of emulsions and capsules having a high degree of droplet-size uniformity, obtained with low mechanical stress input [14–16]. Therefore, the application of membrane emulsiﬁcation extended to various ﬁelds, such as drug delivery, biomedicine, food, cosmetics, plastics, chemistry, and some of these applications are now being developed at the commercial level. Their scale vary from large plants in the food industry, to medium-scale use in the polymer industry, and to laboratory-bench scale in biomedicine. In this chapter, the experimental and theoretical bases as well as the applications of the technology will be discussed.

21.2 Membrane Emulsification Basic Concepts

21.2 Membrane Emulsification Basic Concepts

Emulsions and suspensions are key systems for advanced formulations in various industrial sectors. Membrane emulsiﬁcation is a relatively new technology in which membranes are not used as selective barriers to separate substances but as microstructures to form droplets with regular dimensions, that is, uniform or controlled droplet-size distribution (Figure 21.1). Membrane emulsiﬁcations can be generally distinguished in (Figure 21.2): (i) direct membrane emulsiﬁcation (DME), in which the disperse phase is directly fed through the membrane pores to obtain the droplets, and (ii) premix membrane emulsiﬁcation, in which a coarse premixed emulsion is pressed through the membrane pores to reduce and to control the droplet sizes. In general, in the direct membrane emulsiﬁcation, the disperse phase is pressed through a microporous membrane and droplets are formed at the opening of the pore on the other side of the membrane, which is in contact with the continuous phase. Here, droplets that reach a critical dimension can detach either for spontaneous deformation or are sheared by the continuous phase ﬂowing parallel to the surface. In the former case, the driving force for the droplet formation is the surface free-energy minimization, that is, the droplet is formed by spontaneous deformation tending to form a sphere. For example, in quiescent conditions the droplets are formed by means of this mechanism. In the latter case, the shearing stress generated by the continuous phase is the driving force of the droplet detachment. For example, in the crossﬂow membrane emulsiﬁcation (CDME) and stirred membrane emulsiﬁcation droplets are formed by this mechanism. In the premix emulsiﬁcation the basic mechanism for the droplet formation is different from the direct emulsiﬁcation. In fact, in this case the predominant formation mechanism is the droplet disruption within the pore. Both direct and premix emulsiﬁcation can be obtained with a continuous phase ﬂowing along the membrane surface (i.e., crossﬂow, stirring) (Figure 21.2(b)). However, it is important to distinguish between the droplet-formation mechanism and the macroscopic operation procedure. In other terms, often, in the literature, the

Figure 21.1 Schematic representation (a) of membrane emulsification, where the membrane works as a high-throughput device to form droplets with regular dimensions; (b) photo of an o/w emulsion

j465

j 21 Membrane Emulsification: Principles and Applications

466

Figure 21.2 Schematic drawing of membrane emulsification: (a) mechanisms (b) operation procedures.

crossﬂow term is used to indicate that the continuous phase is ﬂowing along the surface, but this does not guarantee that the shear stress is the driving force for the droplet detachment, as long as the appropriate conditions are not veriﬁed. The membrane emulsiﬁcation can be considered as a case of microdevice emulsiﬁcation process [17, 18] in which the porous membrane is used as microdevices. Membrane emulsiﬁcation carried out in quiescent conditions is also referred to as static membrane emulsiﬁcation, while membrane emulsiﬁcation carried our in moving conditions (either the membrane, i.e., rotating module, or the phase, i.e., crossﬂow) is also referred to as dynamic membrane emulsiﬁcation (Figure 21.2(b)). A peculiar advantage of membrane emulsiﬁcation is that both droplet sizes and size distributions may be carefully and easily controlled by choosing suitable membranes and focusing on some fundamental process parameters reported below. Membrane emulsiﬁcation is also an efﬁcient process, since the energy-density requirement (energy input per cubic meter of emulsion produced, in the range of 104–106 J m3) is low with respect to other conventional mechanical methods (106–108 J m3), especially for emulsions with droplet diameters smaller than 1 mm [1]. The lower energy density requirement also improves the quality and functionality

21.2 Membrane Emulsification Basic Concepts

of labile emulsion ingredients, such as bioactive molecules. In fact, in conventional emulsiﬁcation methods, the high shear rates and the resulting increase of the process temperature have negative effects on shear- or temperature-sensitive components. The shear stresses calculated for a membrane system are much less and it is possible to process shear-sensitive ingredients. The droplet size, its dispersion and the droplet-formation time depend on several parameters: (i) membrane parameters, such as pore-size distribution, pore-border morphology, number of active pores, porosity, wetting property of the membrane surface, (ii) operating parameters, such as crossﬂow velocity (i.e., wall shear stress), transmembrane pressure and disperse-phase ﬂow, temperature, as well as the membrane module used (tubular, ﬂat, spiral-wound); and (iii) phase parameters, such as dynamic interfacial tension, viscosity and density of processed phases, emulsiﬁer types, and concentration. Such quantities combine with different magnitudes, over the ranges of operating conditions, and many of them exhibit coupling effects 4. Moreover, the production of monodisperse emulsions is essentially related to the size distribution of membrane pores and their relative spatial distribution on the membrane surface. It is worth noting that the geometry of the module in which the membrane is located is also an important parameter since it determines in conjugation to the crossﬂow velocity, the wall shear stress (Figure 21.3). Droplet-size distribution and disperse-phase percentage determine the emulsion properties characterizing the ﬁnal formulation for an intended use.

Figure 21.3 Influence of parameters on droplet size and its formation during an emulsification process.

j467

j 21 Membrane Emulsification: Principles and Applications

468

21.3 Experimental Bases of Membrane Emulsification

In this section, an analysis of the experimental observations and empirical correlations related to membrane emulsiﬁcation processes will be illustrated. The theoretical bases that support these results and predict membrane emulsiﬁcation performance will be discussed in the next section. As previously anticipated, the appropriate choice of the membrane dictates the droplet properties. Membranes employed in emulsiﬁcation processes are mainly of inorganic type (ceramic, glassy, metallic), but some examples of polymeric membranes have also been applied. Tables 21.1 and 21.2 summarize some of the most common membranes used in direct and premix membrane emulsiﬁcation, respectively. Most of them have been originally developed for other membrane processes, such as microﬁltration, and adapted in the emulsiﬁcation technology. Nowadays, the growing interest towards membrane emulsiﬁcation is also promoting research efforts in the design and development of membranes speciﬁcally devoted to membrane emulsiﬁcation. Shirasu porous glassy (SPG) membranes were among the ﬁrst membranes speciﬁcally developed for emulsion preparation. SPG membranes are characterized by interconnected micropores, a wide spectrum of available mean pore size (0.1–20 mm) and high porosity (50–60%). Micropore metallic membranes, developed by Micropore Technologies (United Kingdom), are characterized by cylindrical pores, uniform and in a regular array with a signiﬁcant distance between each pore. They are available with pore diameters in the range of 5–20 mm and exhibit very narrow pore-size distribution (Figure 21.4). Membrane-wetting properties may be carefully considered in the membrane selection. In general, the membrane surface where the droplet is formed should not be wetted by the disperse phase. Therefore, a w/o emulsion is prepared using a hydrophobic membrane and an o/w emulsion is prepared using a hydrophilic membrane. On the other hand, w/o and o/w emulsions were successfully prepared using pretreated hydrophilic and hydrophobic membranes, respectively. The pretreatment basically consisted in absorbing the continuous phase on the membrane surface so that to render the membrane nonwetted by the disperse phase [14, 23, 25]. The presence of emulsiﬁer in the disperse phase represents another strategy that permits the preparation of emulsions with a membrane wetted by the disperse phase. The dispersion of droplet diameter mainly depends upon the membrane pore. In general, a linear relationship between membrane pore diameter (Dp) and droplet diameter (Dd) has been observed, especially for membranes with pore diameters larger than 0.1 micrometer. In these cases, linear coefﬁcients varying between 2–10, depending on the operating conditions and emulsion composition, have been obtained [3, 23, 27]. Figure 21.5 summarizes the behavior of the mentioned relationships for different emulsion systems. In general, for a certain emulsion type and in comparable operating conditions, the lower the pore size the lower the droplet size. Fluid-dynamic operating conditions, such as axial or angular velocity (i.e., shear stress that determines drag force value) and transmembrane pressure (that determines disperse-phase ﬂux, for a given disperse-phase viscosity and membrane

Tubular

Tubular

Tubular

Hollow ﬁber

Alumina/hydrophilic

Zirconia/hydrophilic

Polymeric membrane Polyamide/hydrophilic

e ¼ 60%

e ¼ 35%

—

—

Cylindrical pore/e ¼ 61%

10; 50 kDa (NMWCO)a

0.1

0.5–0.8

0.5–0.2

0.1–0.8

0.68

0.6

Forschstung Institut Berghof, (Germany)

Societe des Ceramiques Techniques (France)

Pall-Exekia (France)

Westfalia Separator Membraﬂow (Germany) Societe des Ceramiques Techniques (France) Pall-Exekia (France)

Lab-made

Lab-made

ISE Chemical Industries Co. Ltd (Japan)

Tubular or disk Cylindrical pore/50% < e < 60% 0.2–1.36; 10.2–16.2

Pore diameter Membrane (lm) producer

SPG Technology Co., Ltd. (Japan)

Pore geometry/porosity (e)

Tubular or disk Tortuous, interconnected 0.1–20 cylindrical pore/50% < e < 60%

Ceramic membrane Mullite ceramic/hydrophilic Disk

Silica glass/hydrophobic

MPG/hydrophilic

Porous glass membrane SPG/hydrophilic or hydrophobic

Membrane material/wetting Membrane property configuration

Table 21.1 List of most common membranes used in direct membrane emulsification.

o/w emulsions

o/w emulsions

o/w, w/o/w emulsions o/w emulsions

w/o emulsions

w/o emulsions

o/w and w/o emulsions

o/w, w/o emulsions

Application

(Continued)

[29]

[27, 28]

[5, 26–28]

[25]

[24]

[21–23]

[16, 19–21]

Reference

21.3 Experimental Bases of Membrane Emulsification

j469

Nominal molecular weight cutoff.

—

Polypropylene/hydrophobic Hollow ﬁber

a

— 0.4

0.2–03

10

5% < e < 20%

Disk

Polycarbonate/hydrophilic

Cellulose acetate/hydrophilic Disk

0.5–50

e ¼ 79%

Disk

PTFE/hydrophobic

Wuppertal (Germany)

Advantec Toyo (Japan)

ISOPORE, Nihon Millipore Co. (Japan)

Japan Goretex Co. (Japan)

Pore diameter Membrane (lm) producer

Pore geometry/porosity (e)

(Continued)

Membrane material/wetting Membrane property configuration

Table 21.1

[31]

[30]

Reference

w/o emulsions

[14]

w/o/w emulsions [32]

Application

470

j 21 Membrane Emulsification: Principles and Applications

Tubular

Tubular

Porous glass membrane SPG

Ceramic membrane Alumina/hydrophilic

Cellulose acetate Polyamide

ﬂat ﬂat

Polymeric membrane Polycarbonate ﬂat

Membrane configuration

Membrane material

Deadend Deadend

Deadend, multipass

0.2, 0.45, 0.8, and 3.0 0.8

0.6, 0.8, and 3.0

3.2, 4, and 11

10.7

Deadend, multipass

Deadend, multipass

2.7 and 4.2

Pore diameter (lm)

Crossﬂow

Operation mode

Table 21.2 List of most common membranes used in premix membrane emulsification.

w/o/w emulsions o/w emulsions

o/w emulsions

w/o/w emulsions

w/o/w emulsions

o/w emulsions

Application

Millipore corporation (United State) Advantec Toyo (Japan) Whatman Intl. Ltd., Maidstone (England)

Lab-made

SPG Technology Co., Ltd. (Japan)

Membrane producer

[35] [36]

[34]

[33]

[15]

[11]

References

21.3 Experimental Bases of Membrane Emulsification

j471

j 21 Membrane Emulsification: Principles and Applications

472

Figure 21.4 Porous membranes developed for emulsification processes. (a) Shirasu porous glassy membrane (from SPG Technology Co., LTD, Japan), (b) metallic membrane (From Micropore Technologies, United Kingdom).

Figure 21.5 Relationship between membrane pore diameter (Dp) and droplet diameter (Dd) (Data extrapolated from Refs. [3, 23, 27]).

properties), can be properly adapted to tune emulsion properties. The commonly observed behavior of shear stress and disperse-phase ﬂux on Dd/Dp ratio is depicted in Figures 21.6 and 21.7, respectively. The droplet size decreases with increasing shear stress at the membrane surface and decreasing of the disperse-phase ﬂux. However, the latter inﬂuence is less predominant and depends on the dropletformation time, which in turn is strongly affected by the interfacial dynamical tension. If the droplet-formation time is larger than the complete adsorption of the emulsiﬁer (equilibrium interfacial tension) the lower the inﬂuence of the dispersephase ﬂux. Therefore, in appropriate conditions and for emulsions with droplet size above a micrometer (1–50 micrometer, so-called macroemulsions), transmembrane

21.3 Experimental Bases of Membrane Emulsification

Figure 21.6 Relationship between wall shear stress (t) and Dd/Dp (Data extrapolated from Refs. [27, 28]).

Figure 21.7 Relationship between dispersed phase flux ( Jd) and Dd/Dp (Data extrapolated from Refs. [5, 16, 22, 26]).

pressure may inﬂuence the disperse-phase ﬂux, but have little inﬂuence on changing the droplet size. Dynamic interfacial tension, therefore the emulsiﬁer used, and related adsorption kinetics inﬂuence the emulsiﬁcation process. In general, the faster an emulsiﬁer adsorbs to the newly formed interface, the lower the interfacial tension the smaller the droplet produced. Figure 21.8 shows a linear behavior between the Dd/Dp ratio and interfacial tension. The axial velocity affects the droplet size by both inﬂuencing the surfactant mass transfer to the newly formed interface (that speeds up the reduction of the interfacial tension) and the drag force (that pulls droplets away from the pore mouth). When production of submicrometer droplet size is aimed at, the continuous-phase shear stress and disperse-phase ﬂux have to match the need for small droplet (i.e., high shear stress and low disperse-phase ﬂux) with the need for a reliable system productivity (i.e., high disperse-phase ﬂux).

j473

j 21 Membrane Emulsification: Principles and Applications

474

Figure 21.8 Relationship between interfacial tension (g) and Dd/Dp (Data extrapolated from Ref. [26]).

The physical chemical properties of the phases can inﬂuence droplet formation as well as their stability in the bulk. For example, the viscosity of the continuous phase inﬂuences both the shear stress at the membrane wall and the adsorption kinetics of the emulsiﬁer. Concerning thermodynamically unstable emulsions, the creation of new interfaces from the disruption of the disperse phase increases the free energy of the system, which tends to return to the original two separate systems. Therefore, the use of emulsiﬁer is necessary not only to reduce the interracial tension, but also to avoid the coalescence and the formation of macroaggregates thanks to electrostatic repulsion between adsorbed emulsiﬁer. 21.3.1 Post-Emulsification Steps for Microcapsules Production

In this paragraph, a description of postemulsiﬁcation steps needed to complete the preparation of microcapsules is reported. Microencapsulation can be described as the formation of small, coated particles loaded with a solid, a liquid, a solid–liquid dispersion, gas or solid–gas dispersion (Figure 21.9). The concept of microencapsulation originated in the 1950s and provided the means by which ink formulations used in carbonless copy paper are packaged. This application has been most successful and has led to the development of other applications like the production of microcapsules for thermal printing, optical recording, photocopy toners,

Figure 21.9 Schematic drawing of a microcapsule.

21.3 Experimental Bases of Membrane Emulsification

Figure 21.10 Steps involved in the formation of microcapsules.

diazo copying, herbicides, animal repellents, pesticides, oral and injectable pharmaceuticals, cosmetics, food ingredients, adhesives, curing agents, and live-cell encapsulation [38]. The size of these capsules may range from 100 nm to about 1 mm. Therefore, they can be classiﬁed as nano-, micro- and macrocapsules, depending on their size. The ﬁrst commercial microcapsules were made by Green with a process called complex coacervation [37]. Since then, other methods for preparing microcapsules have been developed of which some are based exclusively on physical phenomena. Some utilize polymerization reactions to produce a capsule shell. Others combine physical and chemical phenomena. But they all have three main steps in common. The steps of the microencapulation preparation are schematically depicted in Figure 21.10. In the ﬁrst step, a dispersion or emulsion has to be formed, followed by deposition of the material that forms the capsule wall (Figure 21.10, step 2). After solidiﬁcation or crosslinking (step 3) of the droplets prepared, the capsules are isolated in the last step. One of the major problems related to the capsule formation is capsule agglomeration. It involves the irreversible or largely irreversible sticking together of microcapsules that can occur during the encapsulation process and/or during the isolation step. The microencapsulation process can be classiﬁed into two main categories (as deﬁned by Thies [38] and reported in Table 21.3): (a) chemical process and (b) mechanical process.

Table 21.3 List of encapsulation processes (After [38]).

Chemical process

Mechanical process

Complex coacervation Polymer/polymer incompatibility Interfacial polymerization in liquid media In situ polymerization In-liquid drying Thermal and ionic gelation in liquid media

Spray drying Spray chilling Fluidized bed Electrostatic deposition Centrifugal extrusion Spinning disc at liquid/gas or solid gas interface Pressure extrusion or spraying into solvent extraction bath

Desolvation in liquid media

j475

j 21 Membrane Emulsification: Principles and Applications

476

Table 21.4 Commercial encapsulation processes and obtained capsule size (After [38]).

Process

Usual capsule size (lm)

Spray drying In-liquid drying or solvent evaporation Polymer phase separation (coacervation) Rotational suspension separation Fluidized bed (Wurster)

5–5000 <1–1000 20–1000 >50 <100

Capsules produced by a chemical process are formed entirely in a liquid-ﬁlled stirred tank or tubular reactor. Mechanical processes use a gas phase at some stage of the encapsulation process. In Table 21.4 the typical size of capsules produced is identiﬁed by a number of processes that have been commercialized. 21.3.2 Membrane Emulsification Devices

The various membrane emulsiﬁcation procedure can be practised by using appropriate membranes and devices conﬁguration. The crossﬂow membrane emulsiﬁcation can be obtained either with tubular or ﬂat-sheet membranes, which are ﬁxed in appropriate housing modules connected to circuits controlling ﬂuid-dynamic conditions. A schematic drawing of a crossﬂow plant is reported in Figure 21.11. The ﬁgure also illustrates the tubular and ﬂat-sheet membranes and modules. SPG (Japan) and Micropore (UK) were among the ﬁrst companies producing plants for crossﬂow membrane emulsiﬁcation. Figure 21.12 shows pictures of common marketed equipments. Emulsiﬁcation devices where the membrane is immersed in a stirred vessel containing the continuous phase, so as to obtain a batch emulsiﬁcation device operating in deadend emulsiﬁcation mode, have also been developed (Figure 21.13). Both ﬂat-sheet and tubular membranes are used. In this membrane emulsiﬁcation device, the continuous phase kept in motion creates the shear stress at the membrane surface that detaches the forming droplets. In a different operation mode, that is, when the continuous phase is not stirred, droplet formation in quiescent conditions is obtained. Rotating membrane emulsiﬁcation is another type of batch emulsiﬁcation. In this case a tubular membrane immersed in a continuous phase vessel is rotating itself and its angular velocity creates the shear stress at the membrane surface (Figure 21.14). Both crossﬂow and deadend systems can be used in premix and direct membrane emulsiﬁcation. In the crossﬂow premix system the coarse emulsion is diluted by permeation into pure continuous phase/diluted emulsion recirculating at the lowpressure side of the membrane. In the deadend system the ﬁne emulsion is withdrawn as a product after passing through the membrane, without any recirculation and/or dilution with the continuous phase. In this process, the ﬁne emulsion can

21.3 Experimental Bases of Membrane Emulsification

Figure 21.11 Schematic drawing of a crossflow plant, using either tubular or flat-sheet membranes.

Figure 21.12 Marketed equipments for membrane emulsification. (a) Plant for crossflow membrane emulsification produced by SPG Technologies Co. Ltd (http://www.spg-techno. co.jp/); (b) spiral-wound metallic membrane module produced by Micropore Technologies (http://www.micropore.co.uk/).

j477

j 21 Membrane Emulsification: Principles and Applications

478

Figure 21.13 Emulsification devices where the membrane is immersed in a stirred vessel containing the continuous phase. Transmembrane pressure applied from (a) external or shell side, and (b) internal or lumen side.

Figure 21.14 Rotating emulsification device.

be repeatedly passed through the same membrane a number of times to achieve additional droplet-size reduction and enhance size uniformity (multipass premix membrane emulsiﬁcation). Each type of device has speciﬁc advantages and disadvantages. The batch emulsiﬁcation is suitable for laboratory-scale investigations. The construction of the device is simple and handling during emulsiﬁcation as well as for cleaning. Crossﬂow membrane emulsiﬁcation is used when it is important that a proper adjustment of all process parameters and larger amounts of emulsion have to be produced.

21.4 Theoretical Bases of Membrane Emulsification

A potential disadvantage of crossﬂow direct membrane emulsiﬁcation is the relatively low maximum disperse-phase ﬂux through the membrane (0.01–0.1 m3/m2 h). Membrane, ﬂuid properties, and transmembrane pressure determine the disperse-phase ﬂux through the membrane. The opportune choice of membrane properties permits control of the ﬂux during membrane emulsiﬁcation process to be obtained. Due to the low productivity, that is, long production time, crossﬂow direct membrane emulsiﬁcation is more suitable for the preparation of relatively diluted emulsions with disperse phase content up to 30%. Nevertheless, this process enables very narrow droplet-size distribution to be produced over a wide range of mean droplet size. Crossﬂow premix membrane emulsiﬁcation holds several advantages over crossﬂow direct membrane emulsiﬁcation. In fact, disperse-phase ﬂuxes of the former emulsiﬁcation process are typically above 1 m3/m2 h, which is one to two orders of magnitude higher than the latter. In addition, the mean droplet sizes that can be achieved using the same membrane and phase compositions are smaller. Also, the experimental apparatus is generally simpler and the process is easier to control and operate since the driving pressure and emulsiﬁer properties are not so critical for the successful operation as in crossﬂow direct membrane emulsiﬁcation. One of the disadvantages of premix membrane emulsiﬁcation is a higher droplet polydispersity.

21.4 Theoretical Bases of Membrane Emulsification

From the theoretical point of view the key problem of the membrane emulsiﬁcation is to explain and predict the dependence of the mean droplet diameter, Dd on the aforementioned membrane emulsiﬁcation parameters. Important quantities such as droplet-formation time can thus be successively predicted by the mean droplet diameter and disperse-phase ﬂux. Droplet formation during direct membrane emulsiﬁcation and in particular in crossﬂow emulsiﬁcation has been described using models different in the scale and in the considered mathematical and physical phenomena, such as: (a) balance equations involving global forces, (b) surface free-energy minimization, (c) microscopic modeling using computational ﬂuid dynamics (CFD) and lattice Boltzmann approaches. The global balance models are less accurate than the other methods, however, they are easier to handle and more instructive. The latter feature is crucial to acquire the necessary understanding of the physical causes at the basis of the droplet formation and detachment. The balance methods are versatile and permit analysis of the inﬂuence of many membrane emulsiﬁcation parameters with limited computational time, useful in process optimizations. Starting from these considerations, in this section more attention will be paid to the proposed torque and force balances.

j479

j 21 Membrane Emulsification: Principles and Applications

480

The balances approaches have to necessarily incorporate approximations and fundamental hypotheses, which reduce the prediction capability of the latter. Every hypothesis comes from a postulated droplet-formation mechanism. The formation mechanism, however, depends signiﬁcantly on the mentioned operating, membrane and phase parameters, thus, it is very difﬁcult to ﬁnd one mechanism valid for all possible parameters values. Consequently, more accurate computation procedures, such as the microscopic modeling or methods using the minimization of the droplet surface, are necessary for the detailed description of droplet formation and accurate predictions. 21.4.1 Torque and Force Balances

This section breaks down as follows: ﬁrst, the macroscopic forces acting on the droplet growing at the pore opening will be discussed, then the balance equations where these forces are involved will be dealt with. However, the accurate derivation of these forces is not reported here because it is beyond the aim of this contribution and therefore only the expressions of the forces used in balance equations will be presented. The forces acting on a droplet attached to the pore opening can be conveniently subdivided into perpendicular and parallel direction with respect to the membrane surface. Considering the former case, the Young–Laplace FYL [39] (named also static pressure force), the dynamic lift FDL and buoyancy FBG forces [26] are generally taken into account. They are deﬁned as: FYL ¼

g pD2p Dd

FDL ¼ 0:761

1:5 0:5 tc;s rc D3d mc

1 FBG ¼ pgDrD3d 6

ð21:1Þ

ð21:2Þ ð21:3Þ

where Dp and Dd correspond to the average membrane pore and droplet diameter, respectively, g is the liquid–liquid interfacial tension, while tc,s, rc and mc represent the shear stress, density and viscosity of the continuous phase, respectively. The quantity Dr in Equation 21.3 represents the difference between the continuous- and disperse-phase densities. However, various authors [13, 26] showed that for small pores (e.g., smaller than 2 mm), the FDL and FBG are negligible with respect the FYL. The inertial force deﬁned by the following equation 26: ð ð21:4Þ FI ¼ rd v2m dA ¼ rd Ap v2m Ap

caused by the disperse-phase ﬂow, with mean velocity vm, would be another perpendicular force to consider. Here, Ap is the pore surface. Concerning the nature of this force, recently it has been emphasized [40] that it has a predominantly viscous

21.4 Theoretical Bases of Membrane Emulsification

character rather than inertial. Starting from this observation, a more accurate expression of this hydrodynamic force has been determined. This force is explicitly dependent on the mean velocity of the disperse phase as well as of the pore diameter. The mean velocity of the disperse phase in turn depends on the effective pressure DPeff. Neglecting the pressure drop due to the membrane pore length, DPeff is equal to the difference between the transmembrane pressure and pressure drop necessary to overcome the capillary effect, that is the Laplace pressure [39]. To ensure monodisperse droplets and to avoid jets of the disperse-phase ﬂux, the transmembrane pressure should never be markedly higher than the Laplace critical pressure. In these conditions, the DPeff produces a small mean velocity and a negligible inertial force or hydrodynamic force with respect to FYL and the drag force. The general expression used to consider the drag force FDR [13, 26, 39] due to the continuousphase crossﬂow and parallel to membrane surface is the following: 3 FDR ¼ kx ptc;s D2d 2

ð21:5Þ

where the parameter kx is equal to 1.7 and takes into account the wall correction factor for a single sphere touching an impermeable wall [41]. In Equation 21.5, the approximation vmc (1/2) tc,sDd is adopted, where v is the undisturbed crossﬂow velocity. The shear stress, evaluated at the droplet center, is assumed equal to that at the membrane surface, which is the wall shear stress. Referring to the expression 21.5, two important considerations are necessary. The ﬁrst concerns the disperse-to-continuous viscosity ratio. In fact, in Equation 21.5 is only considered the viscosity of the continuous phase because this expression is Stockss law corrected to account for the interaction with the membrane surface. Nevertheless, the disperseto-continuous viscosity ratio can signiﬁcantly affect the values of the effective drag force [42]. This consideration is connected with the droplet deformability; the solid particle approximation (Stockss law) can be a restricted assumption in direct membrane emulsiﬁcation modeling. The second observation concerning the FDR is connected with the value of the wall correction factor parameter kx. The reported value was obtained considering a solid droplet leaned on a surface, this assumption should be improved in the case of a droplet growing from a pore. Although different FDR are presented in the literature [43], the wall shear stress at the membrane surface always appears explicitly in these expressions. This quantity depends on the membrane geometry and module. For simple modules (e.g., tubular or ﬂat) consolidated expressions of the wall shear stress can be found [39]. For more complex equipments holding the membrane (e.g., rotating or vibrating systems), the evaluation of the shear stress requires more complicated analysis and calculations [43]. A particular consideration is necessary for the vibrating systems. Recently, it has been emphasized [44] that vibrations of the membrane introduces additional inertial and drag forces (secondary drag force) in a direction parallel to the membrane surface. These two forces depend on the excitation amplitude and the excitation frequency. All presented forces are detaching forces. An increase in these forces will decrease the diameter of the droplets. On the contrary, the interfacial force caused by a uniform interfacial tension along the pore border is a holding force; increasing this force will

j481

j 21 Membrane Emulsification: Principles and Applications

482

increase the droplet size. The more simple and common expression of this force is [13, 39, 45] deﬁned as: Fg ¼ pDp g sin q

ð21:6Þ

where q represents the contact angle. The capillary force 21.6 is obtained by the integration along the pore perimeter of the dFg , force acting on dL of the pore border. The force dFg has magnitude gdL and is directed towards the pore. The contact angle q is assumed constant in the integration. This constraint could be a severe approximation. In fact, the droplet, during its formation, could counterbalance the actions due to the continuous ﬂuid crossﬂow and FYL, FDL, FBG, and FI forces by changing its contact angle along the pore border, that is the droplet twists on the surface. The droplet inclination yields an interfacial force sufﬁcient to keep the droplet on the membrane. Thus, the global Fg should be rewritten in order to take into account the change of contact angle on the contact line [46]. This interfacial force can be expressed as sum of the two components Fgi and Fgk. Considering the contact line G of generic size and shape, the interfacial force can be deﬁned according the following two components: ð Fgi ¼ gðM mÞm i dG parallel to the membrane surface G ð Fgk ¼ gðM kÞk k dG perpendicular to the membrane surface G

ð21:7Þ where M and m are the unit vectors, whose directions are indicated in Figure 21.15.

Figure 21.15 Droplet formation at the pore opening. Side view with vectors indicating the unit vectors, M and m at the pore perimeter and the advancing (qa) and receding (qr) contact angles. VRd represents the crossflow velocity of continuous phase at height equal to droplet radius, and vm is the mean disperse-phase velocity.

21.4 Theoretical Bases of Membrane Emulsification

Concerning the shape of G, it should be noted that the droplet contact line does not necessarily coincide with the pore border. In fact, depending on the afﬁnity between the membrane and the disperse phase, the droplet can spread around the pore. Using the deﬁnition for the contact angle cos q ¼ M m; Equation 21.7 yields the components of the interfacial tension as a function of the contact angle along the contact line G. The integration in Equation 21.7 can be conveniently carried out by dividing the contact line into four sections: the advancing (Ga) and receding (Gr) portions, along which the contact angles assume the constant values qa and qr, respectively, and the two lines corresponding to the transition zones (TZ) in which the contact angles are not constant. Thus, Equation 21.7 can be expanded in a more easy to handle form as: ð ð ð cos qðGÞm i dG Fgi ¼ g cos qa m i dG þ g cos qr m i dG þ 2g Ga TZ ðGr ð21:8Þ sinqðGÞ dG Fgk ¼ g sin qa Ga þ g sin qr Gr þ 2g TZ

It is worth noting that if qa and qr are equal to the equilibrium contact angle q, then the ﬁrst component becomes zero, whereas the second component reduces to the above Equation 21.6. For more information on the difference between qa, qr and equilibrium contact angle given by the Young equation the reader may refer to the original works [46, 47]. The presented forces are used in different balance equations according to mechanics assumptions. From the mechanics point of view two possible states for an immiscible droplet injected into a liquid continuous phase can be considered: (a) the droplet may maintain a spherical symmetry until it begins its detachment (rigid spherical cap conﬁguration) and here the most appropriate mechanical model would be a torque balance, and (b) a deformed droplet at its base, where a force balance at the pore perimeter would be the most suitable model. Both the torque and force balances can be used to derive equations that will deﬁne the diameter of the droplet before its detachment. To be able to calculate the instant when the droplet has grown sufﬁciently to detach from the pore, Peng and Williams [13] at ﬁrst suggested a torque balance around a pore edge: FDR h ¼ ðFg FYL FBG Þ

Dp 2

ð21:9Þ

in which h is the droplet height from the membrane surface. If the droplet shape is signiﬁcantly deformed towards the membrane then h can be approximated with the pore radius and Equation 21.8 reduces to: FDR ¼ ðFg FYL FBG Þ

ð21:10Þ

In Equations 21.9 and 21.10, the interfacial force 21.6 is used by employing a contact angle equal to p/2. Generally in Equation 21.9 the droplet height is substituted by the droplet radius, Dd/2. These equations permit calculation of which diameter torques in clockwise and anticlockwise directions are balanced, beyond this value the droplet detaches. However, for a short period the droplet still maintains its

j483

j 21 Membrane Emulsification: Principles and Applications

484

connection with the pore through a neck. When this connection is completely broken the droplet detachment is complete 48. Kelder et al. and van Rijn [44, 49] used a force balance equation to predict the ﬁnal droplet diameter, that is the droplet size at the end of its detachment. In their model the droplet is attached at the pore border by a curved neck (similar to a strap). The droplet is not leaned on the pore opening but on the membrane surface. Combining Equations 21.6 and 21.5 yields the following forcebalance equation: FDR ¼ Fg

ð21:11Þ

Although the last two balances are similar, it is worth noting that from the mechanics point of view they are different. In fact, the ﬁrst derives from an approximation on the droplet height, whereas the second derives from a different droplet conﬁguration with respect to the membrane surface. The ﬁrst balances (Equations 21.9 and 21.10) estimate the droplet volume at the beginning of the detachment, whereas the second balance (Equation 21.11) estimates it at the end of the detachment ( ﬁnal droplet dimension). The FDR used by Kelder et al. is acting on the center of the droplet, thus it is not expressed as a function of the wall shear stress. However, in both force balances, the holding force 21.6 is used as the interfacial force. De Luca and Drioli [46] at ﬁrst proposed a balance force model along the droplet contact line G using the interfacial force 21.8 instead of Equation 21.6. Components of the interfacial force 21.8 have to counterbalance both the drag force and the forces in the direction perpendicular to membrane surface (all the detaching forces). The droplet deformability is taken into account through the evaluation of the advancing and receding contact angles along the droplet–pore contact line. The detachment is supposed to occur when the interfacial force at the droplet base is unable to counterbalance, through the droplet inclination, the actions of the detaching forces. The resultant set of force balance equations is: parallel to the membrane surface Fgi ðqa ; qr Þ þ FDR ¼ 0 Fgk ðqa ; qr Þ þ FYL þ FDL þ FBG ¼ 0 perpendicular to the membrane surface ð21:12Þ where the dependence of the interfacial tension force on the contact angles is reported for clarity. This set of equations can be solved at every droplet diameter to ﬁnd the contact angles providing the equilibrium of forces. The solution of Equation 21.12 is the set of qa and qr values for any value of the droplet diameter chosen as parameter. The solution paths are in all cases closed lines lying within a minimum and maximum Dd value corresponding to the initial pore diameter (D0 > Dp) and critical value denoted by Dc (critical droplet diameter), respectively. Since no solution exists for a droplet diameter larger than Dc, then it is concluded that this value has the meaning of droplet diameter corresponding to which the detachment of the droplet starts. In those cases where solution branches are found to be physically unacceptable, Dc is taken to be the smallest diameter corresponding to one of the two contact angles reaching the value of p. It is worth noting that if for particular qa and qr values the ﬁrst equation in the balances 21.12 is not satisﬁed but the second equation could be satisﬁed, the droplet should glide along the surface without

21.4 Theoretical Bases of Membrane Emulsification

detaching from the membrane. Recently, Christov et al. [17] have proposed a balanceforce equation similar to the balance 21.11 but for the droplet detachment in quiescent conditions. In this model the interfacial force 21.6, corrected with the Harkins-Brown factor fd, is counterbalanced by the hydrodynamic force that substitutes the drag force FDR. Using an appropriate expression for the mean velocity of the disperse phase deﬁning the hydrodynamic force, the authors proposed a balance equation between these two forces. It is worth noting that in quiescent conditions and for a spherical disperse/water interface (i.e., a spherical sector), the Young–Laplace force 21.1 is always balanced for a contact angle equal to arcsin (Dp/Dd). Therefore, in this condition the hydrodynamic force could be the driving force for the droplet detachment. The last consideration of this section concerns the coupling effects occurring in membrane emulsiﬁcation. The adsorption of the emulsiﬁer on the droplet interface changes the values of the interfacial tension and consequently the interfacial force [26, 50, 51]. In addition, the equilibrium contact angle qc changes as the emulsiﬁer is adsorbed. As shown by [26, 50, 51] the effects of emulsiﬁer adsorption depend on the ratio between droplet-formation time and emulsiﬁer adsorption rate. If the dropletformation time is large enough to permit a complete adsorption of emulsiﬁer then the equilibrium interfacial tension can be used in the above force expressions. On the contrary, if the droplet-formation process is not large enough then the dynamic interfacial tension function has to be considered as a substitute for the scalar equilibrium interfacial tension. In this case the coupling between the droplet diameter and dynamic interfacial tension has been introduced in the correlated balance equations [50]. 21.4.2 Surface-Energy Minimization

Rayner et al. [52] analyzed the formation mechanism of a droplet from a single pore into a quiescent continuous phase condition evaluating the dimensionless Reynolds, Bond, Weber, and the capillary numbers. The magnitude of these dimensionless numbers was calculated for the following membrane emulsiﬁcation setup: 1 mm pore diameter, 5 mN/m interfacial tension, 8 102 kg/m3 oil density, 5 103 Pa s viscosity and 1 103 m/s disperse-phase velocity and zero continuous-phase velocity. The above dimensionless numbers indicate that the interfacial force 21.6 absolutely dominates the emulsiﬁcation process; the hydrodynamic force due to the dispersephase ﬂow is negligible and the drag force absent. In other terms, the holding interfacial force always balances the other involved forces until the complete detachment of the droplet due to a spontaneous deformation. Thus, the balance (Equations 21.9–21.12) is not predictive in this case because the deformation of the droplet is not included. Starting from this consideration, Rayner et al. [52] analyzed the spontaneoustransformation-based (STB) droplet-formation mechanism from the point of view of the surface Gibbs free energy with the help of the Surface Evolver code. Rayner et al. estimated the difference of surface free energy of the droplet before (E1) and after (E2)

j485

j 21 Membrane Emulsification: Principles and Applications

486

its detachment, E2–E1. When E1 is larger than E2 then the spontaneous droplet formation begins. Surface Evolver code was used to evaluate E1 energy. In particular, for an assigned disperse–continuous–surfactant interfacial tension, Surface Evolver, for each droplet volume attached at the pore perimeter, yields the droplet surface with the minimum energy among the possible surfaces related with the given volume. This minimization must respect important constraints, that is the geometry of the pore border and the equilibrium contact angle that is set as a contact energy around the pore. Once the minimum E1 energy is found, the E2 energy have to be evaluated. The E2 value is the free energy of the detached droplet having the same volume of the attached droplet plus the energy of the pore opening. The maximum stable droplet volume (MSV) is the volume of the attached droplet just before the STB droplet formation takes place, that is when E1 > E2. Surface Evolver also gives the possibility to ﬁnd the maximum stable droplet volume thought the Hessian eigenvalues analysis. The occurrence of negative eigenvalues corresponds to the point at which the E2–E1 difference becomes negative. The MSV yields an estimation of the largest droplet that should be formed. It is well known that there is a certain volume of disperse phase remaining attached at the pore. Rayner et al. estimated this remaining volumes using the pressure pinch constraint principle. This principle is based on the division of the droplet MSV into two parts having relative sizes that show an equal Laplace pressure across the surface of both volumes. Using this principle the authors yielded an estimation of the droplet diameters in quiescent conditions and for very low disperse-phase ﬂow. The Rayner et al. approach estimates the maximum dimension achievable for the droplet. The Surface-Evolver-based simulations also showed that for pores with aspect ratio (maximum to minimum length) greater than three the necking formation should occur inside the pore. On the contrary, when the aspect ratio is smaller than three the droplet necking took place outside the membrane pore. The same authors [53] used this approach to analyze the effect of the dynamic surfactant coverage on the ﬁnal droplet size coupled to the expansion rate of the continuous/disperse interface. They found that the dynamic surfactant coverage has a signiﬁcant inﬂuence on the ﬁnal droplet size during the analyzed membrane emulsiﬁcation process. 21.4.3 Microfluid Dynamics Approaches: The Shape of the Droplets

In the microﬂuid dynamics approaches the continuity and Navier–Stokes equation coupled with methodologies for tracking the disperse/continuous interface are used to describe the droplet formation in quiescent and crossﬂow continuous conditions. Ohta et al. [54] used a computational ﬂuid dynamics (CFD) approach to analyze the single-droplet-formation process at an oriﬁce under pressure pulse conditions (pulsed sieve-plate column). Abrahamse et al. [55] simulated the process of the droplet break-up in crossﬂow membrane emulsiﬁcation using an equal computational ﬂuid dynamics procedure. They calculated the minimum distance between two membrane pores as a function of crossﬂow velocity and pore size. This minimum distance is important to optimize the space between two pores on the membrane

21.4 Theoretical Bases of Membrane Emulsification

surface (i.e., the membrane porosity) in order to avoid droplet coalescence on the surface. They characterized the mechanism of the droplet formation (droplet shape, pressure drop through the pore, etc.) occurring for the assigned conditions. Quite recently, Kobayashi etal. [56]carried outa numericalinvestigationontheformation of an oil droplet in water from straight-through microchannels (MC) with an elliptic crosssection and in quiescent conditions. In particular, these CFD simulations demonstrated that the neck formation considerably depends on the aspect ratio of the elliptic MC. Continuous outﬂow of the oil phase from the channel opening was observed for elliptic MCs below a threshold aspect ratio between 3 and 3.5. On the contrary, a droplet with neck inside the membrane pore was found for a droplet formed in the elliptic MCs exceeding the above threshold aspect ratio. This result is in agreement with the conclusion found by Rayner et al. [52] reported above. Cristini and Tan recently reviewed numerical simulations of droplet dynamics in complex ﬂows [57]. The computational ﬂuid dynamics investigations listed here are all based on the socalled volume-of-ﬂuid method (VOF) used to follow the dynamics of the disperse/ continuous phase interface. The VOF method is a technique that represents the interface between two ﬂuids deﬁning an F function. This function is chosen with a value of unity at any cell occupied by disperse phase and zero elsewhere. A unit value of F corresponds to a cell full of disperse phase, whereas a zero value indicates that the cell contains only continuous phase. Cells with F values between zero and one contain the liquid/liquid interface. In addition to the above continuity and Navier–Stokes equation solved by the ﬁnite-volume method, an equation governing the time dependence of the F function therefore has to be solved. A constant value of the interfacial tension is implemented in the summarized algorithm, however, the diffusion of emulsiﬁer from continuous phase toward the droplet interface and its adsorption remains still an important issue and challenge in the computational ﬂuiddynamic framework. The CFD procedures brieﬂy presented are a valid tool for an accurate in-silico analysis of the droplet-formation mechanisms occurring under various membrane emulsiﬁcation parameters. This knowledge can be used in the formulation and validation of the basic assumptions characterizing the aforementioned balance models. Validated computation ﬂuid dynamic models are useful to design optimal membranes and related equipments. In other words, the CFD procedure can be used for in-silico experiments avoiding expensive experimental trial and error tests. Nevertheless, it is worth noting that the CFD simulations for membrane emulsiﬁcation processes are time-consuming tasks. This aspect can be restrictive if many in-silico experiments have to be carried out. Although the CFD procedures give useful information on the droplet break up, not all phenomena are modeled on a solid physical basis, which can result in ambiguous conclusions as in the case for the modeling the contact line dynamics. Other CFD approaches that do not use the VOF procedure (e.g., level-set procedure) should be taken into account. However, this approach in the membrane emulsiﬁcation is still at an early stage of development. Lattice Boltzmann (LB) is a relatively new simulation technique and it represents an alternative numerical approach in the hydrodynamics of complex ﬂuids. The LB method can be interpreted as an unusual ﬁnite-difference solution of the continuity

j487

j 21 Membrane Emulsification: Principles and Applications

488

and Navier–Stokes equation and it is suitable for modeling of multiphase systems. The LB is based on hypothetical ﬂuid particles (packages of ﬂuid) moving and colliding on a lattice according to the kinetic gas theory. One of the most important reasons why the LB algorithm works well for multiphase problems is that the interfaces appear and move automatically during the simulation. By contrast to the mentioned CDF method, it is not necessary to track the interface explicitly. In addition, the implementation of complex wetting conditions (e.g., patterned surfaces) and the dynamics of contact line turns out to be more simple and accurate with respect to the traditional CFD approaches. Moreover, the diffusion and dynamic adsorption of emulsiﬁers during the droplet formation is another aspect that can be correctly treated in a LB framework. In general, the LB simulations for membrane emulsiﬁcation processes are less intensive (time consuming) with respect to the analogous CFD ones. Although the LB methodology has found applications in different areas of ﬂuid dynamics, including simulations of ﬂows in porous media and droplet formation in liquid–gas systems [58], at the moment only the work of van der Graaf et al. [18] is addressed to the droplet formation from a T-shaped microchannel in a liquid–liquid system. It is worth noting that the T-shaped microchannel geometry was approximated as a model of a membrane pore. Although the premix membrane emulsiﬁcation can yield larger ﬂuxes with respect to direct membrane emulsiﬁcation neither methods using surface-energy minimization nor microﬂuid dynamics approaches have been until now reported on the theoretical treatment of the premix membrane emulsiﬁcation.

21.5 Membrane Emulsification Applications 21.5.1 Applications in the Food Industry

Emulsions play an important role in the formulation of foods, that is, o/w emulsions are used for preparation of dressings, artiﬁcial milks, cream liqueurs, and w/o emulsions are used in the production of margarines and low-fat spreads. Food products must have appropriate texture properties. For example, it is important that mayonnaise products have thick and creamy textures, but not too high a viscosity. The rheological properties depend on their composition, such as the concentration of oil droplets or the concentration of thickening agents. The development of membrane emulsiﬁcation technologies permits production of small and uniform droplets and capsules, using mild conditions of temperature, shear stress and pressure. Furthermore, they are able to produce stable droplets with reduced stabilizers content, which will contribute to the manufacturing of improved food products with low-fat content. In this context, the Morinaga Milk Industry (Japan) developed and commercialized a very low fat spread using membrane emulsiﬁcation technology [59, 60]. The advantages in the production of low-fat spreads made the process one of the ﬁrst

21.5 Membrane Emulsification Applications

large-scale applications of membrane emulsiﬁcation. A w/o emulsion using a MPG hydrophilic membrane, previously treated with the oil phase, has been prepared by crossﬂow membrane emulsiﬁcation. The resulting product was stable and free from aqueous phase separation, tasted smooth and melted extremely easily in the mouth. For practical applications in the food industry, where large-volume production is conducted, it is especially important to obtain high disperse-phase ﬂux. Abrahamse et al. [8] reported on the industrial-scale production of culinary cream. In this study they evaluated the required membrane area for different types of membranes: an SPG membrane, an a-Al2O3 membrane and a microsieve ﬁlter. The requirements for culinary cream production were: a droplet size between 1 and 3 mm and a production volume of 20 m3/h containing 30% disperse phase. They concluded that to produce large quantities of monodisperse emulsions the most suitable was a microsieve with an area requirement of around 1 m2. Katoh et al. [3] prepared w/o emulsions composed of salt solution, polyglycerin polyricinolate (PGPR) at 2%wt and corn oil. It has been proven that the dispersephase ﬂux was increased 100-fold using a hydrophilic membrane pretreated by immersion in the oil phase. This made the membrane emulsiﬁcation system practical for large-scale production of a w/o emulsion in food application. Double emulsions are also very useful for food application. Sensitive food materials and ﬂavors can be encapsulated in w/o/w emulsions. Sensory tests have indicated that there is a signiﬁcant taste difference between w/o/w emulsions and o/w emulsions containing the same ingredients, and that there is a delayed release of ﬂavor in double emulsions [61]. W/o/w or o/w/o multiple emulsions having a concentrated aqueous-soluble ﬂavor or a concentrated oil-soluble ﬂavor encapsulated in the internal phase can be prepared. Food products obtained with these particulates exhibit enhanced ﬂavor perception and extended shelf-life [62]. 21.5.2 Applications in the Pharmaceutical Industry

Among the applications of membrane emulsiﬁcation, drug-delivery system (DDS) is one of the most attractive ﬁelds. W/o/w emulsions have been prepared to transport and deliver anticancer drug [4, 63–65]. The emulsion was directly administered into the liver using a catheter into the hepatic artery. In this way, it was possible to suppress the strong side effects of the anticancer drug and also concentrate the dosage selectively to focus on the cancer. The clinical study showed that the texture of the cancer rapidly contracted and its volume decreased to a quarter of its initial size. Composite emulsion as carrier of hydrophilic medicine for chemotherapy was prepared by adding albumin to the internal water phase and lecithin or cholesterol to the oil phase, thus obtaining a water-in-oil emulsion. This emulsion was then pressed through Millipore membrane into an external water phase to form a w/o/w multiple emulsion. Its advantages are high size uniformity and high storage stability [66]. Nakajima et al. referred to membrane emulsiﬁcation as a method to make functional ethanol-in-oil-in-water (e/o/w) emulsions. These e/o/w emulsions are suitable to encapsulate functional components that have a low water and oil solubility

j489

j 21 Membrane Emulsification: Principles and Applications

490

while being soluble in ethanol. An example is taxol, which is an anticancer terpenoid [67]. Vladisavljevic et al. reported on the production of multiple w/o/w emulsions for drug-delivery systems by extruding a coarse w/o/w emulsion ﬁve times through a SPG membrane [68]. Several studies also reported on the preparation of biodegradable polymer microcapsules to be used as drug-delivery systems due to their biodegradable nature and proven biocompatibility. The biopolymers employed are mainly poly(lactide) (PLA) [69], poly(lactic-co-glycolic acid) (PLGA) [70–74], chitosan [75, 76], and calcium alginate [77]. Such polymers have been applied for encapsulating proteins and peptides used as prophylactic and therapeutic agents in biomedical ﬁelds. So far, the delivery route is injection, which not only causes distress and inconvenience to patients, but also induces unstable curative effective and side effects. This is due to the fact that the drugs have to be given frequently, resulting in rapid increase and decrease of drug concentration in blood [75]. Therefore, a sustained delivery system for proteins and peptides is necessary not only for injection administration but also for developing an oral-administration system. The use of microspheres as a controlled release system is one of the prospective methods. In fact, it may prevent encapsulated drugs from degradation by proteolytic enzymes, prolong its half-life and improve its bioavailability in vivo by controlling the release rate of the drug from the microspheres. The preparation of monodisperse hydrogel microspheres, such as poly-acrylamide-co-acrylic acid, poly(N-isopropylacrylamide-co-acrylic acid), has been performed for drug devices thanks to their biocompatibility [77, 79]. The average diameters of the microspheres were dependent on the pore sizes (from 0.33 to 1.70 mm) of SPG membranes used in the preparation procedure. Solid lipid nanoparticles (SLN) have also been introduced as an alternative to solid particles, emulsions and liposomes in cosmetic and pharmaceutical preparations. Charcosset et al. reported the use of membrane emulsiﬁcation for the production of SLN [80]. The lipid phase was pressed through the membrane pores into the aqueous continuous phase, at a temperature above the melting point of the lipid. The SLN are then formed by the following cooling of the preparation to room temperature. The lipids remain solid also at body temperature. The inﬂuence of process parameters on the size and the lipid-phase ﬂux was investigated. The membranes used were supplied by Kerasep ceramic membranes with an active ZrO2 layer on an Al2O3TiO2 support. Three different microﬁltration membranes were investigated: 0.1, 0.2, and 0.45 mm mean membrane pore size. It was shown that SLN nanoparticles could be prepared with a liquid-phase ﬂux between 0.15 and 0.35 m3/h m2 and mean SLN size between 70 and 215 nm. 21.5.3 Applications in the Electronics Industry

The membrane emulsiﬁcation technique is also employed for the preparation of microspheres starting from monomers such as methacrylates (methylmethacrylate, cyclohexyl acrylate, etc.), polyimide prepolymer, styrene monomer [81], and so on.

21.5 Membrane Emulsification Applications

The occlusion of functional materials such as the polyimide prepolymer (PIP) in uniform polymer particles, can ﬁnd promising applications in sophisticated electronic devices such as adhesive spacers of liquid-crystal panel boards (after a minor screening process), adhesives or insulators for microtip circuits, and so forth. Omi et al. [82] showed that about 30% occlusion of polyimide prepolymer (diphenilmethane-4,40 -bis-allylnagiimide, BAN-I-M) was accomplished in the preparation of polymer particles composed of styrene, various acrylates and a crosslinking agent (ethyleneglycol dimethacrylate, EGDMA) via the emulsiﬁcation technique with SPG membrane. Particles with a diameter of 6–12 micrometers were prepared. The presence of acrylates and EGDMA was essential to obtain stable lattices of styrenebased copolymers that occlude BAN-I-M. However, the presence of acrylates with longer side chains, BA and 2EHA, promoted the inclusion of BAN-I-M. In particular, the latter yielded a stable latex occluding 100% of the initial BANI-M without the crosslinking matrix and using octyl alcohol as a stabilizing agent. The lattices without a crosslinking network resulted in an excellent adhesive ability. Guang Hui Ma et al. [83] prepared microcapsules with narrow size distribution, in which hexadecane (HD) was used as the oily core and poly(styrene-co-dimethylamino-ethyl metahcrylate) [P(st-DMAEMA] as the wall. The emulsion was ﬁrst prepared using SPG membranes and a subsequent suspension polymerization process was performed to complete the microcapsule formation. Experimental and simulated results conﬁrmed that high monomer conversion, high HD fraction, and addition of DMAEMA hydrophilic monomer were three main factors for the complete encapsulation of HD. The droplets were polymerized at 70 C and the obtained microcapsules have a diameter ranging from 6 to 10 mm, six times larger than the membrane pore size of 1.4 mm. Furthermore, such monomers can be readily emulsiﬁed by dissolving in volatile solvents such as methylene chloride and chloroform. Uniform polylactide particles, and composite polystyrene (PST) and polymethyl methacrylate (PMMA) particles were produced by solvent evaporation [84–86]. 21.5.4 Other Applications

Membrane emulsiﬁcation has also been applied for the preparation of oil-in-water emulsions to be used in cosmetics and/or dermatology, in particular for the treatment, protection, care, cleaning and make-up of the skin, mucous membranes and hair. The emulsion was composed by oil-phase globules having an average size less than 20 mm; it was prepared by direct membrane emulsiﬁcation through a porous hydrophilic glass membrane having an average pore size ranging from 0.1 to 5 mm and preferably from 0.3 to 3 mm [87]. The technology also represents a suitable strategy for the preparation of multiphase reaction systems that use phase transfer (bio)catalysts. Giorno et al. [88] reported on the use of membrane emulsiﬁcation to distribute lipase from Candida rugosa at the interface of stable oil-in-water emulsions. The enzyme itself was used as a surfactant. Shirasu Porous Glassy (SPG) membranes having a nominal pore

j491

j 21 Membrane Emulsification: Principles and Applications

492

diameter of 0.1 mm were used to prepare emulsions. Emulsions with more than 90% of organic droplets of 1.6 (0.40) mm were obtained. The methodology allowed preservation of the catalytic performance of the biocatalyst as well as optimal enzyme distribution at the interface of stable, uniform and small oil droplets to be achieved. Applications in the chemical ﬁeld, include extrusion of an oil phase containing a photographic hydrophobic material through a microporous membrane into water [89] and emulsiﬁcation of low-viscosity parafﬁn wax in water [90]. The polyurethane (PU) can be considered an environment-friendly material because the urethane bond resembles the amide bond, which implies possible biodegradability. It can be used in various elastomer formulations, paints, adhesives for polymers and glass, and artiﬁcial leather as well as in biomedical and cosmetic ﬁelds. Polyurethane spheres were prepared from 20/40% of PU prepolymer solution in xylene [91]. PU droplets were formed in water with the SPG membrane of different pore size (1.5–9.5 mm) and then polymerized to form the ﬁnal microspheres. Finally, spherical and solid PU particles of 5 mm were obtained after the removal of the solvent. In another study, Ma et al. reported the formation of uniform polyurethane-vinylpolymer (PUU-VP) hybrid microspheres of about 20 mm, prepared using SPG membranes and a subsequent radical suspension polymerization process [92]. The prepolymers were solubilized in xylene and pressed through the SPG membrane into the continuous phase containing a stabilizer to form uniform droplets. The droplets were left for chain extension at room temperature for some hours with di- and triamines by suspension polymerization at 70 C for 24 h. Solid and spherical PU-VP hybrid particles with a smooth surface and a higher destructive strength were obtained. Ha et al. [93, 94] prepared monodisperse polymer microspheres from 1 to 40 mm in diameter for medical diagnostic tests, as chromatography column packing and as calibration standards. The work deals with the synthesis of large and uniform poly (butadiene-styrene) latex. The ceramic SPG membrane, with a pore diameter of 1.6 mm, was employed. The uniform particle sizes were in the diameter range of 4–6 mm. Westover et al. 95 prepared lightly crosslinked nitrated poly(4-hydroxystyrene) microspheres for pH sensors. The microspheres were produced using SPG membranes followed by suspension polymerization and they showed diameters between 1 and 2 micrometers. Figoli et al. [96, 97] reported the preparation of polymeric capsules combining the phase-inversion technique with the membrane process. Polyetheretherketone (PEEKWC) capsules of different size (300–800 micrometer) and morphology (asymmetric with a porous or dense layer) have been prepared. The SEM pictures of the prepared PEEKWC capsules are shown in Figure 21.16. The capsules can ﬁnd application both in chemical and in food packaging ﬁelds [98]. Another ﬁeld where emulsions are likely to become imperative is the production of fuel [99]. Simple and multiple emulsions represent alternative fuels for diesel engines to both increase combustion efﬁciency and reduce particulate emission. Considering the enormous volume of diesel that is being consumed today, a replacement of just a fraction of regular diesel by diesel emulsion could be of considerable interest to the surface chemistry community. Until now, diesel emulsions were prepared by

21.6 Conclusions

Figure 21.16 SEM pictures of the crosssection of the PEEKWC capsules prepared by the phase-inversion technique using a film with a pore size of 550 mm [97].

conventional emulsiﬁcation methods but it is expected that the membrane emulsiﬁcation technique will also become attractive for this application.

21.6 Conclusions

Membrane emulsiﬁcation, a technology that ﬁrst appeared in the early 1990s, is gaining increasing attention with many applications being explored in various ﬁelds. Nowadays, it can be considered at a developing/exploiting stage with a signiﬁcant involvement of industrial and academic research effort. Many studies have been carried out, especially from the experimental point of view whereas from the theoretical point of view the knowledge is not accordingly advanced. In this chapter, a description of membrane emulsiﬁcation basic concepts, empirical correlations, theoretical studies, as well as most common applications have been discussed. Many patents have been applied for, especially in Japan, which currently holds more than 60% of worldwide applications, in Europe and USA. Main drivers for membrane emulsiﬁcation development include high product quality – especially when labile molecules are involved, precise deﬁnition of dropletsize distribution, low energy input, equipment modularity and easy scale-up, and low equipment footprint. Challenges in this ﬁeld include the need for higher productivity, membranes and modules speciﬁcally designed for the emulsiﬁcation process, modules construction standardization, and design of innovative intensiﬁed processes.

j493

j 21 Membrane Emulsification: Principles and Applications

494

References 1 Nakashima, T., Shimizu, M. and Kukizaki, M. (1991) Membrane Emulsiﬁcation Operation Manual, 1st edn, Industrial Research Institute of Miyazaki Prefecture, Japan. 2 Nakashima, T. and Shimizu, M. (1991) Key Engineering Materials, 61/62, 513–516. 3 Katoh, R., Asano, Y., Furuya, A., Sotoyama, K. and Tomita, M. (1996) Journal of Membrane Science, 113 (1), 131–135. 4 Nakashima, T., Shimizu, M. and Kukizaki, M. (2000) Advanced Drug Delivery Reviews, 45, 47–56. 5 Schr€oder, V. and Schubert, H. (1999) Colloid and Surfaces, 152 (1), 103–109. 6 Joscelyne, S.M. and Tr€agårdh, G. (2000) Journal of Membrane Science, 169, 107–117. 7 Charcosset, C., Limayem, I. and Fessi, H. (2004) Journal of Chemical Technology and Biotechnology, 79, 209–218. 8 Abrahamse, A.J., van der Padt, A. and Boom, R.M. (2004) Journal of Membrane Science, 230, 149–159. 9 Vladisavljevi,c G.T. and Williams, R.A. (2005) Advances in Colloid Interface Science, 113, 1–20. 10 Lambrich, U. and Schubert, H. (2005) Journal of Membrane Science, 257, 76–84. 11 Suzuki, K., Shuto, I. and Hagura, Y. (1996) Food Science and Technology International Tokyo, 2 (1), 43–47. 12 Kawakatsu, T., Kikuchi, Y. and Nakajima, M. (1997) Journal of the American Oil Chemists Society, 74, 317–321. 13 Peng, S.J. and Williams, R.A. (1998) Transactions of IChemE, 76, 894–901. 14 Vladisavljevi,c G.T., Tesch, S. and Schubert, H. (2002) Chemical Engineering and Processing, 41, 231–238. 15 Vladisavljevi,c G.T., Shimizu, M. and Nakashima, T. (2004) Journal of Membrane Science, 244 (1–2), 97–106. 16 Vladisavljevi,c G.T. and Schubert, H. (2003) Journal of Membrane Science, 223, 15–23.

17 Christov, N.C., Danov, K.D., Danova, D.K. and Kralchevsky, P.A. (2008) Langmuir, 24, 1397–1410. 18 van der Graaf, S., Nisisako, T., Schr€oen, C.G.P.H., van der Sman, R.G.M. and Boom, R.M. (2006) Langmuir, 22, 4144–4152. 19 Cheng, C.J., Chu, L.Y. and Xie, R. (2006) Journal of Colloids and Interface Science, 300, 375–382. 20 Sotoyama, K., Asano, Y., Ihara, K., Takahashi, K. and Doi, K. (1999) Journal of Food Science, 64 (2), 221–215. 21 Vladisavljevi,c G.T., Kobayashi, I., Nakajima, M., Williams, R.A., Shimizu, M. and Nakashima, T. (2007) Journal of Membrane Science, 302, 243–253. 22 Scherze, I., Marzilger, K. and Muschiolik, G. (1999) Colloids and Surface B, 12, 213–221. 23 Mine, Y., Shimizu, M. and Nakashima, T. (1996) Colloids and Surfaces B: Biointerfaces, 6 (4), 261–268. 24 Fuchigami, T., Toki, M. and Nakanishi, K. (2000) Journal of Sol-Gel Science and Technology, 19, 337–341. 25 Wu, J., Jing, W., Xing, W. and Xu, N. (2006) Desalination, 193, 381–386. 26 Schr€oder, V., Behrend, O. and Schubert, H. (1998) Journal of Colloid and Interface Science, 202, 334–340. 27 Joscelyne, S.M. and Tr€agårdh, G. (1999) Journal of Food Engineering, 39, 59–64. 28 Berot, S., Giraudet, S., Riaublanc, A., Anton, M. and Popineau, Y. (2003) Transactions of IChemE, 81, 1077–1082. 29 Giorno, L., Mazzei, R., Oriolo, M., De Luca, G., Davoli, M. and Drioli, E. (2005) Journal of Colloid and Interface Science, 287, 612–623. 30 Yamazaki, N., Yuyama, H., Nagai, M., Ma, G.H. and Omi, S. (2002) Journal of Dispersion Science and Technology, 23 (1–3), 279–292. 31 Kobayashi, I., Yasuno, M., Iwamoto, S., Shono, A., Satoh, K. and Nakajima, M.

References

32

33

34

35

36

37

38

39

40

41 42 43

44

45

(2002) Colloids and Surface A, 207, 185–196. Shima, M., Kobayashi, Y., Fujii, T., Tanaka, M., Kimura, Y., Adachi, S. and Matsuno, R. (2004) Food Hydrocolloids, 18 (1), 61–70. Correia, L.A., Pex, P.A.C., van der Padt, A. and Poortinga, A.T. (2004) Presented at 8th International Conference on Inorganic Membrane (ICIM 9), 2004, Cincinnati, USA, 18–22 July. Park, S., Yamaguchi, T. and Nakao, S. (2001) Chemical Engineering Science, 56 (11), 3539–3548. Shima, M., Kobayashi, Y., Fujii, T., Tanaka, M., Kimura, Y., Adachi, S. and Matsuno, R. (2004) Food Hydrocolloids, 18, 61–70. Ribeiro, H.S., Rico, L.G., Badolato, G.G. and Schubert, H. (2005) Journal of Food Science, 70 (2), E117–E123. Kondo, A. (1979) History and classiﬁcation of microencapsulation, in Microcapsule Processing and Technology (ed. J. Wade Van Valkenburg), Marcel Dekker, New York, NY, Chapter 4. Thies, C. (1996) A survey of microencapsulation processes, in Microencapsulation Methods and Industrial Applications (ed. S. Benita), vol. 17, Marcel Dekker, New York, pp. 1–21. Wang, Z., Wang, S., Volker, S. and Schubert, H. (2000) Chinese Journal of Chemical Engineering, 8, 108–112. Danov, K.D., Danova, D.K. and Kralchevsky, P.A. (2007) Journal of Colloid and Interface Science, 316(2), 844–857. ONeil, M.E. (1964) Mathematika, 11, 67–74. Ken, H.J. and Chen, P.Y. (2001) Chemical Engineering Science, 56, 6863–6871. Kosvintsev, S.R., Gasparini, G., Holdich, R.G., Cumming, I.W. and Stillwell, M.T. (2005) Industrial & Engineering Chemistry Research, 44, 9323–9330. Kelder, J.D.H., Janssen, J.J.M. and Boom, R.M. (2007) Journal of Membrane Science, 304, 50–59. Chatterjee, J. (2002) Advances in Colloid Interface Science, 98, 265–283.

46 De Luca, G. and Drioli, E. (2006) Journal of Colloid and Interface Science, 294, 436–448. 47 De Luca, G., Di Maio, F.P., Di Renzo, A. and Drioli, E. (2008) Chemical Engineering and Processing, 47, 1150–1158. 48 Xu, J.H., Luo, G.S., Chen, G.G. and Wang, J.D. (2005) Journal of Membrane Science, 266, 121–131. 49 van Rijn, C.J.M. (2000) Nano and Micro Engineered Membrane Technology in Membrane Science and Technology Series, vol. 10, Amsterdam, Elsevier. 50 De Luca, G., Sindona, A., Giorno, L. and Drioli, E. (2004) Journal of Membrane Science, 229, 199–209. 51 van der Graaf, S., Schr€oen, C.G.P.H., van der Sman, R.G.M. and Boom, R.M. (2004) Journal of Colloid and Interface Science, 277, 456–463. 52 Rayner, M., Tr€agårdh, G., Tr€agårdh, C. and Dejmek, P. (2004) Journal of Colloid and Interface Science, 279, 175–185. 53 Rayner, M., Tr€agårdh, G. and Tr€agårdh, C. (2005) Colloids and Surface A: Physicochemical and Engineering, Aspects 266, 1–17. 54 Ohta, M., Yamamoto, M. and Suzuki, M.M. (1995) Chemical Engineering Science, 50, 2923–2931. 55 Abrahamse, A.J., van der Padt, A., Boom, R.M. and de Heij, W.B.C. (2001) AIChE Journal, 47 (6), 1285–1291. 56 Kobayashi, I., Mukataka, S. and Nakajima, M. (2004) Langmuir, 20, 9868–9875. 57 Cristini, V. and Tan, Y.-C. (2004) Lab on a Chip, 4 (257), 257–264. 58 Kalarakis, A.N., Burganos, V.N. and Payatakes, A.C. (2003) Physical Review E, 67, 016702-1–016702-8. 59 Okonogi, S., Kato, R., Asano, Y., Yuguchi, H., Kumazawa, R., Sotoyama, K., Takahashi, K. and Fujimoto, M. (1994) US5279847. 60 Okonogi, S., Kumazawa, R., Toyama, K., Kato, M., Asano, Y., Takahashi, K. and Fujimoto, M. (1992) JP4258251. 61 van der Graaf, S., Schro€en, C.G.P.H. and Boom, R.M. (2005) Journal of Membrane Science, 251, 7–15.

j495

j 21 Membrane Emulsification: Principles and Applications

496

62 Gaonkar, A.G. (1994) US 5332595. 63 Higashi, S., Shimizu, M. and Setoguchi, T. (1996) Colloids and Surface B, 6, 261–268. 64 Higashi, S., Shimizu, M., Nakashima, T., Iwata, K., Uchiyaemotherapyma, F., Tateno, S., Tamura, S. and Setoguchi, T. (1995) Cancer, 75, 1245–1254. 65 Higashi, S. and Setoguchi, T. (2000) Advanced Drug Delivery Reviews, 45, 57–64. 66 Guanghui, M., Hui, S., Zhiguo, S. and Lianyan, W. (2005) CN1600295. 67 Nakajma, M., Nabetani, H., Ichikawa, S. and Xu, Q.Y. (2003) US6538019. 68 Vladisavljevic, G.T., Shimizu, M. and Nakashima, T. (2006) Journal of Membrane Science, 284, 373–383. 69 Liu, R., Ma, G.H., Wan, Y.H. and Su, Z.G. (2005) Colloid and Surface B: Biointerfaces, 45, 144–153. 70 Ito, F. and Makino, K. (2004) Colloid and Surface B: Biointerfaces, 39, 17–21. 71 Omi, S., Katami, K., Yamamoto, A. and Iso, M. (1994) Journal of Applied Polymer Science, 51, 1–11. 72 Shiga, K., Muramatsu, N. and Kondo, T. (1996) Journal of Pharmacy and Pharmacology, 48, 891–895. 73 Costa, M.S. and Cardoso, M.M. (2006) Desalination, 200, 498–500. 74 Shiga, K., Muramatsu, N. and Kondo, T. (1996) Journal of Pharmacy and Pharmacology, 48, 891–895. 75 Wang, L.Y., Ma, G.H. and Su, Z.G. (2005) Journal of Controlled Release, 106, 62–75. 76 Wang, L.Y., Gu, Y.H., Zhou, O.Z., Ma, G.H., Wan, Y.H. and Su, Z.G. (2006) Colloid and Surface B: Biointerfaces, 50, 126–135. 77 Fuchigami, T., Toki, M. and Nakanishi, K. (2000) Journal of Sol-Gel Science and Technology, 19, 337–341. 78 Nagashima, S., Koide, M., Ando, S., Makino, K., Tsukamoto, T. and Ohshima, T. (1999) Colloids and Surfaces A: Physiochemical and Engineering Aspects, 153, 221–227. 79 Makino, K., Agata, H. and Ohshima, H. (2000) Journal of Colloid and Interface Science, 230, 128–134.

80 Charcosset, C., El-Harati, A. and Fessi, H. (2005) Journal of Controlled Release, 108, 112–120. 81 Dowding, P.J., Goodwin, J.W. and Vincent, B. (2001) Colloids and Surfaces A: Physiochemical and Engineering Aspects, 180, 301–309. 82 Omi, S., Matsuda, A., Imamura, K., Nagai, M. and Ma, G.H. (1999) Colloids and Surface A: Physiochemical and Engineering Aspects, 153, 373–381. 83 Ma, G.H., Su, G.Z., Omi, S., Sundberg, D. and Stubbs, J. (2003) Journal of Colloid and Interface Science, 266, 282–294. 84 Ma, G.H., Nagai, M. and Omi, S. (1999) Journal of Colloid and Interface Science, 214, 264–282. 85 Ma, G.H., Nagai, M. and Omi, S. (1999) Colloids Surfaces A: Physiochemical and Engineering Aspects, 153, 383–394. 86 Muramatsu, N. and Kondo, T. (1995) Journal of Microencapsulation, 12, 129–136. 87 Roulier, V. and Quemin, E. (2000) WO0021491. 88 Giorno, L., Piacentini, E., Mazzei, R. and Drioli, E. Journal of Membrane Science, 317 (1–2), 19–25. 89 Kiyoshi, E. (1999) JP11242317. 90 Aryanti, N., Williams, R.A., Hou, R. and Vladisavljevic, G.T. (2006) Desalination, 200, 572–574. 91 Yuyama, H., Yamamoto, K., Shirafuji, K., Nagai, M., Ma, G.H. and Omi, S. (2000) Journal of Applied Polymer Science, 77, 2237–2245. 92 Ma, G.H., An, C.J., Yuyama, H., Su, Z.G. and Omi, S. (2003) Journal of Applied Polymer Science, 89, 163–178. 93 Ha, Y.K., Song, H.S., Lee, H.J. and Kim, J.H. (1999) Colloids and Surface A: Physiochemical and Engineering Aspects, 162, 289–293. 94 Ha, Y.K., Song, H.S., Lee, H.J. and Kim, J.H. (1998) Colloids and Surface A: Physiochemical and Engineering Aspects, 145, 281–284. 95 Westover, D., Seitz, W.R. and Lavine, B.K. (2003) Microchemical Journal, 74, 121–129.

References 96 Figoli, A., De Luca, G., Lamerata, F. and Drioli, E. (2006) Desalination, 199, 115–117. 97 Figoli, A., De Luca, G., Longavita, E. and Drioli, E. (2007) Separation Science and Technology, 42, 2809–2827.

98 Figoli, A., De Luca, G. and Drioli, E. (2007) Italian Journal of Food Science, ISSN 11201770, XI, 90–96. 99 Nakajima, N., Fujiwara, M., Maeda, D. and Watanabe, K. (2006) JP2006182890.

j497

j499

22 Membrane Contactors in Industrial Applications Soccorso Gaeta

Membrane contactors are devices that selectively allow mass transfer between two different streams (gas/liquid or liquid/liquid) that are in contact, without mixing, at the membrane interface. Membrane contactors can be used in several industrial areas: liquid/liquid extraction, gas absorption and stripping, biotechnology applications, pharmaceutical applications, wastewater treatment, metal-ion extraction, the electronics industry, the automobile industry, the food industry, air dehumidiﬁcation, membrane distillation, membrane crystallizers [1], puriﬁcation of ﬂue gases (CO2, SO2, H2S, etc.) [2], and microclimate control (relative humidity, odors, etc.). They are used as end-of-pipe technology but also integrated in the process to recover products. The driving force for the mass transfer usually is a (partial) pressure gradient, a temperature gradient, or a concentration gradient at the membrane interface. In this chapter only membrane contactors systems working with speciﬁc extractants to capture selected molecules will be described. In Figure 22.1 the separation process is schematically described. The two phases are separated by a hydrophobic microporous membrane that acts as an interface between two streams and controls the mass transfer. The two streams (liquid or gas) ﬂow tangentially to the membrane: one stream contains the liquid or gas to be treated and the other stream contains an extractant active towards the speciﬁc molecules that need to be separated from the ﬁrst stream. The hydrophobic membrane does not allow the liquid water-based stream to pass through the membrane; while the extractant contained in the same liquid stream captures at the membrane interface a target gas or vapor molecule contained in the second (gas or liquid) stream. Thus, the membrane does not act as selective media, it just acts as an interface between the two streams. The selectivity is guaranteed by the extractant, which is characterized by a very high afﬁnity for the target molecule. The extractant must be selected very carefully in order to accommodate the process needs. Speciﬁcally, it must be characterized by the following characteristics: high activity towards the molecules to be captured; low energy requirement for regeneration; nontoxic; noncorrosive; low volatile; low viscosity; low cost.

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 22 Membrane Contactors in Industrial Applications

500

Figure 22.1 Schematic representation of a hydrophobic membrane used as interface between a gas and a liquid stream in a membrane contactor.

Membranes are polymeric microporous materials in hollow-ﬁber or ﬂat-sheet conﬁgurations. The membrane properties control the contactor and the membrane contactor system performance and economy. The most important membrane properties are: . . . .

low resistance to mass transfer: this contributes in obtaining high ﬂow rate, hydrophobicity stable with time, high value of water breakthrough: this allows operation at higher partial pressure difference between the two sides of the membrane, cost low enough to have membrane contactor systems competitive with traditional equipment.

However, the engineering optimization of the equipment and of the processes is also very important, that is, all auxiliary pieces of equipment must be properly designed or selected in order to minimize parasitic costs; if not properly selected, auxiliary equipment costs can contribute strongly to the total cost. The experimental results reported in this chapter are related to membrane contactors manufactured by GVS S.P.A by using microporous polymeric ﬂat-sheet

22 Membrane Contactors in Industrial Applications

Figure 22.2 Picture of a membrane contactor housing 1 m2 of membrane.

membranes and cassette-type membrane contactors. The ﬂat conﬁguration of contactors was selected to keep the pressure drops at both sides of the membrane in the contactor as low as possible and to have the possibility to use a large variety of membranes; in fact, the ﬂat membranes commercially available are more numerous than available hollow-ﬁber membranes. The possibility to keep pressure drops low is very important in most applications in order to integrate membrane contactors in existing plants without changing existing auxiliary equipments: pumps, fans, heat exchangers. This contributes to minimizing assembling time and equipment cost. In Figure 22.2 the picture of a membrane contactor manufactured by GVS and used for the demonstration tests reported in this chapter is shown. The mechanism of the capture of a speciﬁc gas molecule by the extractant is also schematically illustrated. The molecule is captured by the extractant without any contact between the stream to be treated and the extractant stream. While the gas or liquid polluted stream passes through the contactor ﬂowing tangentially to one side of the membrane, the gas molecule to be captured passes through the pores of the membranes and is captured at the other side of membrane interface by the extractant solution. Periodically, the extractant is regenerated to release the absorbed molecules and to be reused in the system. In Figure 22.3 a picture of various membrane contactors housing 1 and 3 m2 of membrane is shown. The modules were developed by GVS in partnership with other institutions in the framework of the several projects partially funded by the European Union: FP6ECintegratedprojectUltraLowCO2 Steelmaking — ContractNo.515960; P6 EC integrated project CO2, From Capture To Storage — Contract No. 502686; FP6 LIFE Project Novel Technology to Reduce Greenhouse Gas Emissions Contract No. LIFE 05 ENV/IT/000876. Also, part of the work was funded by the Italian government in the framework of the project PRIITT 3.1 A 09.02.2004 - Project No. 171.

j501

j 22 Membrane Contactors in Industrial Applications

502

Figure 22.3 Picture of two membrane contactors each housing 1 m2 of membrane (right) and 1 contactor housing 3 m2 of membrane (left).

These ﬂat-type membrane contactors have been used in several demonstration tests carried out in the framework of the previously cited European projects or independently by GVS in collaboration with selected industrial end-users. The efforts have been concentrated on two sectors: (1) air dehumidiﬁcation (for short- and medium-term industrial applications), (2) capture of CO2 from ﬂue gases of steel and power plants (for long-term industrial applications). In this chapter the results related to the dehumidiﬁcation of air in refrigerated trucks, the dehumidiﬁcation of air in refrigerated storage rooms, and the capture of CO2 from ﬂue gases of steel industry are described. A good review of industrial application is reported in [3]. In Table 22.1 the characteristics of a membrane contactor housing 1 m2 of ﬂat membrane and used for the demonstration tests related to air dehumidiﬁcation are Table 22.1 Characteristics of the membrane contactor used to

dehumidify air and housing 1 m2 of membrane area. Dimension, mm Membrane area, m2 Speciﬁc surface, m2/m3 Pressure drop gas side, Pa Water vapor ﬂow rate, g/h m2 Water breakthrough - desiccant side, m H2O Thermal resistance, C

251 · 380 · 88 1 120 <200 >1000 >5 max 90

22 Membrane Contactors in Industrial Applications

Figure 22.4 Schematic diagram showing how membrane contactors systems operate.

reported. Dimensions also include connections. The contactors are made of reinforced polypropylene (except the membrane); this suggest to keep the working temperature lower than 90 C. Alternative materials (such as, for example, polyamides) can be used if higher thermal resistance is required. They can easily be assembled in series or parallel conﬁgurations. In Figure 22.4 a schematic diagram showing how membrane contactors systems operate is illustrated. In the contactor the pollutant is captured at the membrane interface; in the stripper, which could also be a membrane contactor, the exhaust extractant solution is regenerated and the pollutant released. The regenerated extractant solution is thereafter returned to the membrane contactor. In Figure 22.5 the general layout of a system using membrane contactors is described. An industrial system comprises a membrane contactor, a regenerator, a liquid desiccant (LD) solution, two pumps to circulate the desiccant in the membrane contactor and in the regenerator, one or two heat exchangers to preheat the diluted desiccant before the regeneration step and to cool the concentrated desiccant before entering the membrane contactor. In the membrane contactor the target molecules are absorbed. In the regenerator the desiccant is regenerated and the absorbed molecule is discharged. This is the main energy input and energy of low quality (from waste heat, natural gas, solar collectors, etc.) must be used when possible.

j503

j 22 Membrane Contactors in Industrial Applications

504

Figure 22.5 General layout of a membrane contactors system.

The advantages over traditional stripping technologies can be summarized as follows: .

there is no mixing between streams, therefore there is zero carryover,

.

it is possible to independently control the ﬂow rates of the two streams,

.

the systems are very compact and the weight of the equipment is lower than currently used absorbers or strippers,

.

high contact area per unit of volume of the system and precise identiﬁcation of the interfacial area,

.

limited pressure drop at both streams sides,

.

no channeling, foaming or by-pass phenomena occur,

.

the noise level is limited,

.

high recovery factors are possible,

.

limited fouling of membranes,

.

systems are manufactured with plastics, corrosion is limited,

.

membrane contactors can be scaled up easily because they are modular system working in series and parallel: more contactors in series increase the separation efﬁciency, more contactors in parallel increase the capacity of the system. By using modular cassettes it is possible to design a system of any required capacity.

However, industrial applications have been limited mostly because these new membrane-based systems have not been proven extensively at the industrial level and their advantages have not been quantiﬁed in industrial terms, as is the case for traditional consolidated technologies. This fact has discouraged industries from applying these systems in large plants. Therefore, demonstration tests are needed in order to be able to fully exploit the commercial potential of the membrane contactor technology.

22.1 Air Dehumidification: Results of Demonstration Tests with Refrigerated Storage Cells

22.1 Air Dehumidification: Results of Demonstration Tests with Refrigerated Storage Cells and with Refrigerated Trucks

To speed up in the short and medium term the exploitation of the technology, in the last 5 years GVS has carried out several projects to demonstrate the environmental, technical, social and economic viability of processes to dehumidify air based on membrane contactors. The objective is to quantitatively show that by using membrane contactors in refrigerated trucks and storage cell systems it is possible to save energy, decrease time and frequency of defrosting cycles, minimize ice formation on (packed) goods, improve safety of people working in the area, facilitate movement of automated or mechanical equipment, and improve the impact on environment. In fact, by using membrane contactors air at atmospheric pressure can be dehumidiﬁed without cooling, that is, independently from temperature; therefore, these systems have the potential to save energy when compared to traditional vaporcompression cycles. In Figure 22.6 the dehumidiﬁcation process based on membrane contactors is schematically represented. The humidity in the air is captured in the membrane contactor by a liquid desiccant. The liquid desiccant does not pass through the hydrophobic membrane and is stopped at the membrane interface, while the water vapor passes freely through the membrane and is captured by the liquid desiccant at the membrane interface. The driving force for the process is the difference between the vapor pressure in the air and in the desiccant aqueous solution; therefore, air temperature and RH can be controlled independently. Periodically, the desiccant solution is regenerated by stripping off the water absorbed by the desiccant by using a standard stripper or a membrane contactor. The liquid desiccant is a water solution of a highly hygroscopic salt. This technology is important to reduce energy consumption in applications where latent loads are high. Speciﬁcally, the demonstrations have been focused upon the dehumidiﬁcation of air in refrigerated trucks at 5 C and in intermediate cells of refrigerated storage rooms at T ¼ 15 C.

Figure 22.6 Schematic diagram of the dehumidification process.

j505

j 22 Membrane Contactors in Industrial Applications

506

Figure 22.7 Layout of the system: membrane contactor and refrigerated cell of trucks or of storage cells.

The objective of these applications is to minimize the ice formation on evaporator surfaces of cooling equipment, on packaging, on movable parts of equipment, and on ﬂoors. This would allow less use of energy, more comfortable working places, less maintenance, and minimized loss of products. A demonstration unit able to treat 600 m3/h of air has been built and delivered to the demonstration site. In Figure 22.7 the layout of the system (membrane contactor and refrigerated cell of trucks or of storage cells) is described. The chamber was equipped with two refrigerating units working with R404A (reference refrigerant in industry for such applications). The unit was instrumented to measure temperature and pressure at different points of the refrigerating circuit, the refrigerant ﬂow rate, the energy consumption of the compressor, of the evaporator and of the condenser. On the air side calibrated thermocouples to measure temperature at different locations in the cell; two capacitive hygrometers on the evaporator; a humidity controller designed to deliver a maximum vapor ﬂow rate of 10 kg/h have been installed. A liquid ﬂowmeter to measure the volumetric ﬂow rate of the aqueous desiccant solution at the membrane contactor outlet was used. All instruments are connected to a data logger. In Figure 22.8 the layout of the membrane contactor system is described. As liquid desiccant a CaCl2 solution has been used. The demonstration runs lasted more than 450 hours.

22.2 Refrigerated Storage Cells

Figure 22.8 Membrane contactor and its auxiliary equipment.

22.2 Refrigerated Storage Cells

For the refrigerated storage cells application the intermediate cell (IC) application has been selected, because it has emerged that it is more convenient and effective to stop the water vapor ﬂow rate in the intermediate cell at 15 C before it enters the refrigerated storage cell at 25 C. In fact, the energy required to capture the water vapor at 25 C is remarkably higher because when the vapor enters the cold cell it must be condensed and transformed in ice at 25 C (using energy). Therefore, it is better to capture the vapor in the IC. In Figure 22.9 the general layout of the humidity control in the IC in order to reduce the humidity to the low-temperature storage cell (SC) is reported.

Figure 22.9 Schematic diagram of the system studied.

j507

j 22 Membrane Contactors in Industrial Applications

508

The air temperature of the intermediate cell is assumed to be equal to 15 C. This intermediate cell is dehumidiﬁed by the membrane contactor system placed at the evaporator inlet. The dehumidiﬁed air is driven to an air curtain at the top of the door separating the intermediate cell from storage cell. The heat load of the system and the air circulation ratio are assumed to be constant. The temperature of CaCl2 solution at the inlet of the membrane contactor is assumed to be equal to the indoor IC air temperature. The following parameters have been measured: . . . . . . . . . . . . . . . . . . . . . .

outside temperature ¼ 25 C, outside R.H. ¼ 70%, temperature of intermediate cell (IC) ¼ 15 C, temperature of the storage cell (SC) ¼ 30 C, ratio SC volume/IC volume ¼ 3, sensible heat load of IC ¼ 4.8 W/m3, sensible heat load of SC ¼ 1.6 W/m3, air renewal outside air-IC in 12 h working time ¼ 1 h1, air renewal outside air-IC in no-working time ¼ 0.05 h1, air renewal SC-IC in 12 working time ¼ 0.08 h1, air renewal SC-IC in no-working time ¼ 0.003 h1, air curtain efﬁciency between SC and IC ¼ 0.4, contactor membrane area ¼ 8 103 m2/m3 (IC), air ﬂow rate through the contactor ¼ 1.6 m3/h m3(IC), saturated CaCl2 ﬂow (IC temperature) ¼ 1.6 kg/h m3(IC). Door IC – outdoor H ¼ 4 m, W ¼ 3 m, no-working period (12 h) ne ¼ 0.05 [1/h], working period (12 h), 30 opening; Dt ¼ 360 s. Door IC – SC H ¼ 2.5 m, W ¼ 3 m, no-working period (12 h) nnif ¼ 0.005 [1/h], working period (12 h). In these conditions a total energy saving of about 15% can be achieved.

22.3 Refrigerated Trucks

Applications involving transport of fresh refrigerated goods at 5 C have been selected. In fact, these types of refrigerated transports are more demanding in terms of energy because these goods (fruit and vegetables) are generally not packed and the rate of water evaporation is quite high: this generates high energy consumption and spoilage of goods. On the contrary, generally in trucks transporting frozen products (at 25 C) the goods are packed in plastics that prevent evaporation. Therefore, the energy consumption is quite low and the application of the membrane contactors is not very suitable. In Figure 22.10 and in Figure 22.11 the layout of the transport of fresh perishable food at þ5 C is described. The air is dehumidiﬁed in the

22.3 Refrigerated Trucks

Figure 22.10 Scheme of the system considered for refrigerated trucks applications.

membrane contactors. A heat exchanger allows the control of the CaCl2 solution temperature in the tank at T ¼ 5 C. The membrane contactor is placed at the inlet of the evaporator of the refrigerating unit and handles 10 – 20% of the total recycled air ﬂow. This value deﬁnes the by-pass factor (BF). In the demonstrations the BF was 80–90%. After dehumidiﬁcation, the air is mixed with the recycled air. In the tests carried out the resulting energy savings are higher than 20% when captured water is more than 0.25 kg/h. In the case of optimized conditions a reduction of CO2 equal to 20% is possible. The surface temperature of the cooling coil has a great inﬂuence on the dehumidiﬁcation process of the recycled air owing to the ice forming on the cooling coils, that is, the lower the coil temperature, the greater the frost formation, reducing the apparent efﬁciency of the membrane contactor system. The higher the latent load, the higher is the effectiveness of the membrane contactor.

Figure 22.11 Overview of a refrigerated truck equipped with MC.

j509

j 22 Membrane Contactors in Industrial Applications

510

In conclusion, the demonstration runs have shown that the membrane contactor dehumidiﬁcation system is effective when the total loads of the refrigerated cell do not require an evaporative temperature of the cooling coil below 10 C. In these conditions it has been shown that energy, operating costs and CO2 emissions can be decreased up to 20% when optimized systems are built. The optimization is related mostly to the selection of adequate pumps, fans, heat exchangers. The system could also be further improved if membrane contactors are used in hybrid systems integrated with vapor compression inverse cycle; in this case the power saving can be as high as 50% [4].

22.4 Capture of CO2 from Flue Gas

GVS has also dedicated important efforts to demonstrate the possibility to apply membrane contactors to capture CO2 from ﬂue gases of steel and power plants. For CO2 capture mostly PTFE membranes are used [5]; as extractants aqueous solutions of different amines are used [6]. In the past, the use of membrane contactors to capture CO2 has been studied by TNO in the Netherlands and by Kvaerner/Gore in Norway and Germany [7]. They have worked with tubular systems. They claim that working with ﬂue gas from a power generation plant can recover 85% of CO2. In the following section of the chapter results related to the recovery of CO2 from ﬂue gases of conventional blast furnaces of steel plants are described. The work was carried out in the framework of the FP6 EC integrated project Ultra Low CO2 Steelmaking—Contract No. 515960. GVS has used cassette-type membrane contactors as reported in the previous sections of this chapter. In Figure 22.12 a

Figure 22.12 Flow sheet of a membrane contactor plant to capture CO2 from flue gases of a steel manufacturing plant (Data developed in the framework of FP6 EC integrated project Ultra Low CO2 Steelmaking—Contract No. 515960.)

22.4 Capture of CO2 from Flue Gas Table 22.2 Design of an industrial plant treating 300 000 m3/h of flue gas.

Gas composition, IN Air ﬂow rate, IN – m3/h CO2 ﬂow rate, IN – m3/h CO2 ﬂow rate, OUT – m3/h C02 absorbed by1 contactor, m3/h m2 Density of CO2, kg/m3 Gas ﬂow rate, IN - 25% CO2, m3/h CO2 ﬂow rate, IN - m3/h CO2 captured, m3/h CO2 noncaptured, m3/h Contactor membrane area, m2 Contactor cost, D /m2 Total contactor cost, D

75% air þ 25% CO2 30 10 4 6 1.8 300 000 75 000 66 000 9000 11 000 48 528 000

schematic diagram of the application studied is reported. CO2 is captured in the membrane contactor by using a 30% water solution of ethanolamine (MEA). The MEA solution is continuously regenerated in a steam-operated stripper. Process conditions have been optimized in order to obtain the best possible efﬁciency and cost. It has been shown that membrane contactors can be advantageously used to capture CO2 from ﬂue gases containing about 25% by volume of CO2 and to obtain in the decarbonated gas maximum 3% of CO2 mole (i.e. 88% capture of CO2). It has been proven that the contactors can capture up to 6 m3/h of CO2 per m2 of membrane. In Table 22.2 results of a design of a potential industrial plant treating 300 000 m3/h of ﬂue gas are reported. The cost of a single contactor (48 D /m2) is based on a GVS evaluation related to the anticipated cost of contactors in the year 2013.

Table 22.3 Cost of the CO2 captured when treating about 300 000 m3/h of flue gas containing 25% volume of CO2.

Cost of equipment

D 1 320 000

Total steam to regenerate MEA (10 D /t steam) Chemicals consumption Manpower (60 kD /man-year) Variable costs/t of CO2 captured Maintenance CO2 recovery CO2 purity (dry basis) Total CO2 captured (density 1.8 kg/m3) Return on investment (R/I) Amortization (15 yr linear) General overhead Total cost CO2 captured

200 t/h 500 kD /y 1 man/y 17 D /t 41 000 D /y 88% 100% 66 000 m3/h (1 040 688 t/y) 12% 7% 1% 18.3 D /t

j511

j 22 Membrane Contactors in Industrial Applications

512

In Table 22.3 the cost to capture CO2 by using a membrane contactor plant treating about 300 000 m3/h of ﬂue gas (25% volume of CO2) is reported. The cost refers only to the membrane contactor part of the schematic plant reported in Figure 22.12. The cost of remaining equipment has not been evaluated because it refers to standard pieces of equipment extensively used in industries and their cost can vary remarkably depending upon speciﬁc factors of different companies. The resulting ﬁgure, equal to 18.3 D /ton of CO2 recovered, is quite interesting and competitive with traditional technologies to capture CO2, such as, for example, direct contact strippers using amines (the estimated cost to capture CO2 in this case is about 40 D /ton) or absorption processes (VPSA) combined with a cryogenic CO2 puriﬁcation unit (the estimated cost to capture CO2 in this case is about 30 D /ton).

References 1 Drioli, E., Curcio, E. and Di Proﬁo, G. (2005) Trans IchemE, Chemical Engineeging Research and Disign, 83, 1–11. 2 Klaassen, R., Feron, P.H.M. and Jansen, A.E. (2005) Trans IchemE, Chemical Engineeging Research and Disign, 83, 234–246. 3 Gabelman, A. and Hwang, S.T. (1999) Journal of Membrane Science, 159, 61–106. 4 Berghero, S., Chiari, A. and Nannei, E. (2006) Proceedings of ESDA2006, Turin Italy, July.

5 Feron, P. and Jansen, A.E. (2002) Separation Puriﬁcation Technology, 27, 231–242. 6 Kim, Y.S. and Yang, S.M. (2000) Separation Puriﬁcation Technology, 21, 101–109. 7 Svendsen, H.F., Hoff, K.A., Poplsteinova, J. and Silva, E.F. (2001) 2nd Nordic Symposium on CO2 Capture and Storage, Goteborg, October 26.

j513

23 Extractive Separations in Contactors with One and Two Immobilized L/L Interfaces: Applications and Perspectives Štefan Schlosser

23.1 Introduction

Partitioning of components between two immiscible or partially miscible phases is the basis of classical solvent extraction widely used in numerous separations of industrial interest. Extraction is mostly realized in systems with dispergation of one phase into the second phase. Dispergation could be one origin of problems in many systems of interest, like entrainment of organic solvent into aqueous rafﬁnate, formation of stable, difﬁcult-to-separate emulsions, and so on. To solve these problems new ways of contacting of liquids have been developed. An idea to perform separations in three-phase systems with a liquid membrane is relatively new. The ﬁrst papers on supported liquid membranes (SLM) appeared in 1967 [1, 2] and the ﬁrst patent on emulsion liquid membrane was issued in 1968 [3]. If two miscible ﬂuids are separated by a liquid, which is immiscible with them, but enables a mass transport between the ﬂuids, a liquid membrane (LM) is formed. A liquid membrane enables transport of components between two ﬂuids at different rates and in this way to perform separation. When all three phases are liquid this process is called pertraction (PT). In most processes with liquids membrane contact of phases is realized without dispergation of phases. Separations in two-phase systems with one immobilized interface(s) are much newer. The ﬁrst paper on membrane-based solvent extraction (MBSE) published Kim [4] in 1984. However, the inventions of new methods of contacting two and three liquid phases and new types of liquid membranes have led to a signiﬁcant progress in the last forty years. Separations in systems with immobilized interfaces have begun to be employed in industry. New separation processes in two- and three-phase systems with one or two immobilized L/L interfaces realized with the help of microporous hydrophobic wall(s) (support) are alternatives to classical L/L extraction and are schematically shown in Figure 23.1. Membrane-based solvent extraction (MBSE) in a two-phase system with one immobilized interface feed/solvent at the mouth of microspores of hydrophobic support is depicted in Figure 23.1a and will be discussed

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j 23 Extractive Separations in Contactors with One and Two Immobilized

514

Figure 23.1 Two- and three-phase systems with one and two immobilized L/L interfaces realized with help of microporous hydrophobic wall(s) (support). (a) Two-phase system with one immobilized interface F/S, (b) Three-phase system with two immobilized interfaces F/M and M/R, organic phase in pores serves as a supported liquid membrane (SLM), (c) Threephase system with two immobilized interfaces F/M and M/R, organic phase in pores and

between fibers serves as a bulk liquid membrane (BLM), (d) Three-phase system with one immobilized interfaces F/M and emulsion of the stripping solution – bulk liquid membrane with emulsion (BLME). F – aqueous feed, HF – hydrophobic microporous hollow fiber, M – membrane phase (organic), R – aqueous stripping solution, S – organic solvent.

in Section 23.3. Processes with two phases and one immobilized interface are de facto not membrane processes because the microporous wall only immobilize an L/L or L/ G interface and has no separation function in a process. A three-phase system with two immobilized interfaces feed/LM and LM/stripping solution is shown in Figure 23.1b, where organic phase soaked in micropores of hydrophobic hollow ﬁber wall serves as a supported liquid membrane (SLM). Pertraction through SLM and its applications will be discussed in Section 23.5. A three-phase system with two immobilized interfaces feed/LM and LM/stripping solution realized in two hydrophobic walls with organic phase in pores and between ﬁbers (support walls). Organic phase serves as a bulk liquid membrane (BLM) as presented in Figure 23.1c. More information on pertraction through this type of BLM can be found in refs. [5–9]. A three-phase system can be made up of one immobilized feed/LM interface in microporous wall and a third phase, the stripping solution, emulsiﬁed in the bulk of organic phase – BLME. A pure membrane (organic) phase is in the pores of the support wall. The aqueous stripping solution cannot enter hydrophobic pores. BLM with emulsion of stripping solution is in the bulk of the

23.1 Introduction

organic phase as shown in Figure 23.1d and will be further designated as BLME. Pertraction through BLME will be discussed in Section 23.4. Pertraction (PT) can be realized through a liquid membrane, but also through a nonporous polymeric membrane that was applied also industrially [10–12]. Apart from various types of SLM and BLM emulsion liquid membranes (ELM) were also widely studied just at the beginning of liquid membrane research. For example, an emulsion of stripping solution in organic phase, stabilized by surfactant, is dispersed in the aqueous feed. The continuous phase of emulsion forms ELM. Emulsion and feed are usually contacted in mixed column or mixer-settlers as in extraction. EML were applied industrially in zinc recovery from waste solution and in several pilotplant trials [13, 14], but the complexity of the process reduced interest in this system. More information on ELM and related processes can be found in refs. [8, 13–16]. A serious problem, which has not been solved up to now, is the short lifetime of SLM. One way to overcome this is application of pertraction with bulk liquid membranes (BLM) shown in Figures 23.1c and d. This is taken into account by a doubled wall and a thicker liquid membrane layer between walls (bulk membrane) in an arrangement as in Figure 23.1c compared with SLM, which results in a higher mass-transfer resistance [17, 18]. In pertraction through BLME (Figure 23.1d) dispersion or emulsion of stripping solution is used, but the advantage of the nondispersive process is lost. In systems inclined to emulsiﬁcation in stripping this may introduce a problem. The mass-transfer resistance will be lower than in the system in Figure 23.1c. Several mechanisms to achieve transport of solute(s) through the L/L interface or through a liquid membrane can be utilized. The separation mechanism could be based on differences in the physical solubility of the solutes or their solubilization in the solvent or reverse micelles or based on the chemistry and rate of chemical or biochemical reactions occurring on L/L interface(s). The complexing or solubilization agent – extractant (carrier in the liquid membrane) forms by a reversible reaction complex(es) or aggregate(s) with the solute, which are soluble in the solvent or membrane. The chemistry of reactive extraction and stripping in MBSE and MBSS, as well as in PT, is identical to the classical solvent extraction or stripping and is presented in several books and papers, for example, [8, 19–21]. Enzymatic reactions on L/L interfaces were employed to achieve separations [22– 25]. For example, the aqueous feed, a mixture of phenylacetic acid (PAA) and 6-aminopenicillanic acid and the stripping phases ﬂow in the lumen of two bundles of hydrophobic hollow ﬁbers with microporous walls [24]. Fibers are immersed in the immiscible organic phase forming a liquid membrane. PAA reacts at the L/L interface with alcohol added to the feed under the catalytic action of enzyme lipase Candida rugosa and the ester formed dissolves in the membrane and is transported through it. On the downstream interface the desertiﬁcation reaction proceeds catalyzed by another lipase enzyme and PAA is deliberated to the stripping solution with a higher pH than in the feed. The second acid is not transported via this mechanism and separation of acids occurs with a separation factor of 10. Formation of hybrid production/biotransformation – separation processes, including extractive processes, could enhance production and is of great interest, as shown in review papers [8, 21, 26–30]. The ﬂowsheet of the extractive fermentation

j515

j 23 Extractive Separations in Contactors with One and Two Immobilized

516

unit with an integrated MBSE and MBSS circuit for recovery of acid(s)/product from the fermentation broth is discussed in Section 23.3 and with integrated pertraction in Section 23.5. The aim of this chapter is to give a short overview of recent trends on application of extractive separations in contactors with one and two immobilized L/L interfaces and their perspectives. Industrial applications and pilot-plant tests will be stressed and selected case studies highlighted.

23.2 Contactors with Immobilized L/L Interfaces

Contactors with ﬂat-sheet and cylindrical walls are used but only hollow ﬁber (HF) contactors in cylindrical modules in several sizes are available commercially [31]. Flatsheet contactors are widely used in analytical chemistry [32–34]. There are two main types of HF contactors, those with parallel ﬂow of phases in ﬁber lumen and in shell or crossﬂow of phases. A HF contactor with crossﬂow of phases is shown in Figure 23.2. More details on their construction and sizes available are presented in the producers web site [31]. HF contactors with planar elements with ﬂowing head of ﬁbers and crossﬂow of one phase in three and more phases contactor have been suggested in a patent [35] and their scheme is shown in ref. [8]. A two-phase HF contactor with planar elements was developed at TNO and tested in pilot plants [36, 37]. Reviews on two-phase HF contactors are presented in refs. [27, 38–40]. Mass-transfer characteristics of twophase contactors are presented in ref. [30]. Three-phase HF contactors for pertraction are described in refs. [6–9, 41]. They are not produced commercially. HF contactors have a large interfacial area per unit volume of the contactor without the requirement of dispergation of one phase, which can be advantageous in systems sensitive to emulgation [42–45]. The volume ratio of phases could be varied practically without limitations. A disadvantage of HF contactors is connected with additional mass-transfer resistance introduced by porous wall(s) immobilizing L/L interface(s). Some problems with swelling of HF and especially of potting material of HF in organic solvents may occur.

Figure 23.2 Hollow-fiber contactor with crossflow of phases (Liqui Cel Extra-Flow, Membrana [31].

23.3 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS)

Modeling of HF contactors is in most papers based on a simple diffusion resistance in series approach. In many systems with reactive extractants (carriers) it could be of importance to take into account the kinetics of extraction and stripping reactions that can inﬂuence the overall transport rate, as discussed in refs. [30, 46]. A simple shortcut method for the design and simulation of two-phase HF contactors in MBSE and MBSS with the concentration dependent overall mass-transfer and distribution coefﬁcients taking into account also reaction kinetics in L/L interfaces has been suggested [47]. A rotary disc (RD) pertractor with BLM was suggested in refs. [48, 49]. In a RD pertractor hydrophilic discs are ﬁxed on a rotating horizontal shaft. The lower parts of the discs are immersed in compartments, which are alternately ﬁlled with the stripping solution and the feed. The remaining parts of the discs, on which ﬁlms of aqueous phases are formed due to rotation, are immersed in the membrane phase. Mass transfer occurs from the feed ﬁlms into the stripping solution ﬁlms through the bulk liquid membrane. Pertraction in RD contactors has been widely studied in the Boyadzhiev group for recovery of organic acids [50–53], antibiotics [54], alkaloids [55–58], biosurfactant [59] and metals [60–64].

23.3 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS)

Membrane-based solvent extraction (MBSE) is a relatively new alternative of classical solvent extraction where mass transfer between immiscible liquids occurs from the L/L interface immobilized at the mouth of pores of a microporous wall, which is not wetted by one of the phases in contact as shown in Figure 23.1a and in more detail in Figure 23.3. Basic information on MBSE is given in refs. [18, 27, 30, 38, 39, 65]. The solvent can be regenerated by membrane-based solvent stripping (MBSS) where the solute is re-extracted into the stripping solution. Another method of regeneration could be distillation of the volatile solvent or solute, and so on, depending on the properties of the system. A schematic ﬂow sheet of the simultaneous MBSE and

Figure 23.3 Detail view of the two-phase system in membranebased solvent extraction (MBSE) in contactor with hydrophobic wall.

j517

j 23 Extractive Separations in Contactors with One and Two Immobilized

518

Figure 23.4 Flowsheet of MBSE with simultaneous regeneration of the solvent by MBSS in HF contactors and recirculation of the solvent to extraction. In both contactors the solvent flows in the shell.

MBSS processes with closed loop of the solvent is shown in Figure 23.4. In this way recovery of the solvent and concentration of the solute can be achieved. Preferable contactors for MBSE and MBSS are hollow-ﬁber contactors, which are discussed in Chapter 2. Filling the pores with gel to protect against leakage of phase through the pores has been suggested in ref. [66]. The functions of contactors in the simultaneous MBSE and MBSS with an arrangement as shown in Figure 23.4, are coupled. They react similarly as a pertractor with a SLM. The differences are only in the overall resistance, which is smaller in the pertractor where there is only one support wall. In addition, it is not necessary to pump the solvent in its circulation loop in PT, as is used in the simultaneous MBSE and MBSS process [30]. On the other hand, in pertraction through SLM its limited lifetime could be problem which is not the case in MBSE and MBSS where it is easy to keep the constant properties of the solvent phase. Extraction into capsules with a solvent, for example, recovery of phenylethanol (a product of phenylalanine bioconversion by yeast) [67] or lactic acid from fermentation broth [68], has attracted interest recently. The polymeric core of the capsule prevents direct contact of the solvent with biomass. This process could be regarded as a batch MBSE. An interesting variant of MBSE with a dual mechanism of separation is extraction from an ion-exchange membrane, which has been suggested by Kedem and Bromberg [69] and Isono et al. [70, 71]. The separation is performed by L/L partitioning and is enhanced by an electrostatic rejection in the ion-exchange membrane. A very interesting separation of biomolecules, for example, proteins and enzymes (BSA, lysozyme), in an aqueous two-phase system (ATPS) realized in hollow ﬁber contactor is suggested in the paper of Riedl and Raiser [72]. Application of ATPS in

23.3 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS)

Figure 23.5 Schematic flowsheet of the fermentation unit with integrated MBSE and MBSS circuit for recovery of acids (product) from the fermentation broth [30].

MBSE opens up an attractive alternative to classical downstream processing operations as chromatography. A schematic ﬂowsheet of the fermentation unit with integrated MBSE and MBSS circuit for recovery of acid(s) product from the fermentation broth is presented in Figure 23.5. Martak et al. [73] ran a semicontinuous fermentation of lactic acid with Rhizopus arrhizus with a periodical bleed and feed operation without a decrease in LA productivity for 152 h. Such a process could be integrated with separation of lactic acid, for example, by MBSE studied in ref. [74] or by pertraction [44, 45]. Recovery of vanilline from a fermentation broth is presented in ref. [75] aiming at formation of an integrated system. A combination of MBSE of phenol from saline solution in HF contactor with bioreactor with Pseudomonas putida to remove phenol is studied in ref. [76]. Modeling and optimization of MBSE and MBSS of a multicomponent metallic solution in HF contactors is discussed in ref. [77]. A short-cut method for the design and simulation of two-phase HF contactors in MBSE and MBSS with the concentration-dependent overall mass-transfer and distribution coefﬁcients taking into account also reaction kinetics was suggested by Kertesz and Schlosser [47]. Comparison of performance of the MBSE and MBSS circuit with pertraction through ELM in case of phenol removal presented Reis [78] and for copper removal Gameiro [79]. An overview of selected papers on recovery/removal of organic compounds by MBSE and/or MBSS is presented in Table 23.1 and for metals in Table 23.2. 23.3.1 Case Studies

The ﬁrst published information on the industrial application of a hybrid system with a HF contactor for production of the drug dilthiazem intermediate was reported by Lopez and Matson [23]. An enzymatic resolution of dilthiazem chiral intermediate is realized in an extractive enzymatic membrane reactor. The enzyme is entrapped in the macroporous sponge part of the hydrophilic hollow-ﬁber membrane made of a

j519

MBSE MBSE, MBSS MBSE, MBSS MBSE, MBSS MBSE, MBSS PT

Processa

MBSE MBSE

Abbreviations used are explained in the list of abbreviations.

a

Aroma compounds (in oil) BSA, lysozym

MBSE MBSE, MBSS Butyric acid MBSE, MBSS MBSE, MBSS Aspartic acid MBSE, MBSS ()-trans-methyl-methoxyphenyl-glycidate in MBSE toluene Penicillin-G MBSE, MBSS MBSE Vanilline MBSE, MBSS Phenol MBSS MBSE, MBSS MBSE combined with bioreactor MPCA MBSE, MBSS DMCCA MBSE Thiols MBSS Aromatic compounds; trichloroethene MBSE Aroma compounds (aqueous) MBSE

Lactic acid

Phenylalanine

Solute

CF HF CF HF CF HF CF HF CF HF PF HF CF HF CF HF PF HF CF HF CF HF PF HF CF HF CF HF PF HF CF HF CF HF CF HF PF HF CF HF

Amberlite LA2/(kerosene, isodecanol) Amberlite LA-2/(Shellsol TK, TBP) butylacetate from organic feed into aqueous NaOH Cyanex 923/isoparaﬁns kerosene TOA/xylene (TOA or Hostarex A327)/(n-alkanes, iso-decanol) from kerosene into aqueous NaOH Feed to reactor; n.a. sunﬂower oil hexane cyclodextrins/water aqueous two-phase system

PF HF PF HF CF HF CF HF CF HF PF HF(PT)

Contactor typea

Aliquat 336/(kerosene, isodecanol) TOMAC/(n-heptane, hexanol) D2EHPA/kerosene D2EHPA/n-alkanes tertiary amines/(n-alkanes, isodecanol, isotridecanol) TOA/xylene (TOA, Aliquat336)/sunﬂower oil TOA/n-alkanes amines/(corn oil, oleyl alcohol) D2EHPA/kerosene Aqueous buffer

Solventa (extractant/diluent)

Table 23.1 Selected papers on recovery/removal of organic compounds by MBSE and/or MBSS.

[95, 96] [30] [97, 98] [99] [37, 100] [101] [102–105] [106] [72]

[90, 91] [92] [75] [93, 94] [78] [76]

[46] [87] [47] [88] [89] [23]

[80, 81] [82] [83–85] [86] [74]

Literature

520

j 23 Extractive Separations in Contactors with One and Two Immobilized

23.3 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS) Table 23.2 Selected papers on recovery and removal of metals by MBSE and/or MBSS.

Metal

Processa

Solventa (extractant/diluent)

Contactor typea

Literature

Cu(II)

MBSE, MBSS MBSE, MBSS

LIX84/n-decane LIX64N/kerosene; Aliquat 336/kerosene D2EHPA/kerosene Cyanex 302/kerosene LIX 54/parafﬁnic solvent (Shellsol T) pure TBP D2EHPA/n-dodecane Cyanex 302/kerosene Aliquat 336/kerosene Aliquat 336/kerosene Aliquat 336/(kerosene, isodecanol) Cyanex 302/kerosene LIX79/n-heptane LIX79/n-heptane di-n-hexyl octanamide/n. alkanes butyl acetate

PF HF CF HF

[107] [108]

CF HF CF HF CF HF

[46] [109] [79]

CF HF CF HF CF HF CF HF CF HF CF HF

[110–112] [113, 114] [17, 115] [116] [117, 118] [119–121]

PF HF, CF HF CF HF CF HF PF HF

[109] [122] [123] [124]

PF HF

[125, 126]

MBSE MBSE MBSE, MBSS Zn(II) Cd(II) Co(II), Ni(II) Cr/VI)

MBSE, MBSS MBSE, MBSS MBSE, MBSS MBSE, MBSS MBSE, MBSS MBSE, MBSS

Cd(II), Hg(I) Ag(I) Au(I) Th(IV)

MBSE MBSE, MBSS MBSE, MBSS MBSE, MBSS

Tl(III)

MBSE

a

Abbreviations used are explained in the list of abbreviations.

polyacrylonitrile copolymer. The enzyme is loaded to the membrane during ultraﬁltration of an aqueous enzyme solution ﬂowing at the beginning in the shell of a HF contactor. A dense layer of the membrane (skin), retaining enzyme, is on the inner side of the hollow ﬁber. After immobilization of the enzyme in the wall, a toluene solution of reactant, racemic ()-trans-methyl-methoxyphenylglycidate, is ﬂowing in the shell. In the ﬁber lumen an aqueous buffer solution with bisulﬁte anion ﬂows countercurrently. The enzymatic desertiﬁcation catalyzed by lipase proceeds on the L/L interface. The deliberated (2S,3R)-methoxyphenylglycidic acid is extracted to the buffer. The required product (2R,3S)-methyl-methoxyphenylglycidate remaining in toluene is an intermediate for dilthiazem synthesis. In the commercial plant with a capacity of 75 tons of drug per year 24 contactors with a surface area of 60 m2 each are installed. An aromatic compound from industrial waste water from a reactor is removed by MBSE in a HF crossﬂow contactor. This installation, with a capacity of 15 m3 h1 of water, went into operation in 1998 [37, 100]. As a solvent is used the feed into the reactor, and in this way pollutant, is recycled into the process. The recovery of phenol from the hydrocarbon fraction with a phenol concentration of 2–4 wt.% by MBSS into an alkali solution has been applied industrially in Poland [93, 94]. The capacity of the plant with two rigs in series, each with 8 crossﬂow HF contactors Liqui Cel 4 2800 (Membrana) connected in parallel is about 650 kg h1. Both the hydrocarbon rafﬁnate, with less than 0.02 wt.% of phenol,

j521

j 23 Extractive Separations in Contactors with One and Two Immobilized

522

and the phenolate concentrate (25–30 wt. %) are recycled back to the technology producing the waste stream. Valuable organic acids of industrial interest can be recovered from aqueous waste solutions, for example, from an enzymatic resolution process. Mass-transfer data for MBSE and MBSS in HF contactors for dimethylcyclopropanecarboxylic acid (DMCCA) [97] and 5-methyl-2-pyrazinecarboxylic acid (MPCA) [95, 96] have been estimated. Processes for recovery of DMCCA from a highly acidic waste solution with pH below 2 containing about 19 kg m3 of DMCCA by MBSE and MBSS have been suggested [98]. A solvent with 0.4 kmol m3 of TOA in n-alkanes (dodecane fraction) was used. A recovery of more than 90% of DMCCA and a concentrate with about 200 kg m3 of DMCCA can be achieved. Based on laboratory data on MBSE and MBSS of MPCA in the crossﬂow HF contactors 2.5 in. 8 in. [95, 96] a production pilot plant was simulated and designed [30]. 0.1 m3 h1 of the aqueous feed with 0.12 kmol m3 MPCA, 1 kmol m3 Na2SO4, and constant pH (2.5) was extracted in a HF contactor by the solvent with 0.4 kmol m3 TOA in xylene. The loaded solvent was regenerated in a stripping HF contactor with NaOH solution with addition of NH4OH. The concentration factor for MPCA in the loaded stripping solution was 10. From simulations of the pilot plant by the short-cut method suggested in ref. [47] it followed that the number of contactors (length of ﬁbers) in MBSE and MBSS is sensitive to the increase in MPCA yield in MBSE above 90%. The velocity of the feed in ﬁber lumen was 1.9 cm s1. The optimum Re number for ﬂow of the solvent in the shell and the approach to equilibrium at the rafﬁnate end of the HF contactor is about 0.2 and 70%, respectively [30]. The number of contactors Liqui Cel 4 2800 (Membrana) in series, as resulted from simulations for the mentioned process data, was found to be 2 for both MBSE and MBSS, which are reasonable numbers. The technological ﬂowsheet of a production pilot plant unit for recovery of MPCA is shown in Figure 23.6. A large effort of the Juelich team has been devoted to development of fermentation–extraction process for production of phenylalanine (Phe) with integrated MBSE in HF contactors [83–85]. 10 v/v% of D2EHPA in kerosene was used as a reactive solvent. Starting from laboratory experiments in separatory funnels through separations of 42-l batches to a fully integrated pilot process with 300 L fermenter working in fed-batch mode of operation [83, 85]. Two HF contactors with crossﬂow of phases and surface areas of ﬁbers 18.6 m2 were used in MBSE and MBSS working in parallel, as in the scheme shown in Figure 23.5. Based on mass-transfer data for MBSE and MBSS of Phe in a HF contactor with a surface area of 1.4 m2 published in ref. [86] a simulation of the pilot plant for recovery of Phe was done [127]. Number of contactors needed for recovery of Phe was estimated for the unit with a feed ﬂowrate of 100 L h1, Phe concentration of 50 mol m3 in the ﬁltered broth, yield of Phe in MBSE 70%, Reshell ¼ 2.0, approach to the equilibrium at the rafﬁnate end of contactor of 60%, and concentration factor of 10 were supposed. The estimated number of contactors of Liqui-Cel type 400 2800 (Membrana, with an effective length of ﬁbers of 0.6 m and surface area of ﬁbers 19.2 m2) in series was 6 in MBSE and 5 in stripping. From simulations it followed that the number of contactors (length of ﬁbers) in MBSE and MBSS is very sensitive to the

23.3 Membrane-Based Solvent Extraction (MBSE) and Stripping (MBSS)

Figure 23.6 Scheme of a continuous pilot plant unit for recovery of MPCA from mother liquor (ML) by MBSE and MBSS in hollow fiber contactors [30]. 1 – container of filtered feed, 2, 8, 10 – polishing (safety) filters, 3a, 3b – HF contactors for MBSE, 4 – static mixer, 5 – container of H2SO4 (pH adjustment), 6a, 6b – HF contactors for MBSS of

the solvent, 7 – container of the regenerated solvent, 9 – container of the stripping solution, 11 – container of NaOH solution, 12 – container of NH4OH solution, C1 to C5 – pumps, pH1, pH2, pH3 – pH sensors, R1, R2 – flowmeters (rotameters), V1 to V5, VP1, VP2 – valves, VR1, VR2 – valves for Dp adjustment.

increase of the Phe yield in MBSE above 0.70 (70%), Figure 23.7. The dependences of the total number of contactors on the concentration ratio in the solvent (actual to equilibrium concentration) at the rafﬁnate end of the MBSE contactor (approach to the equilibrium) exhibit a minimum at about 0.60 (60%) as shown in Figure 23.8. The total number of contactors monotonously decreases with increasing Reynolds

j523

j 23 Extractive Separations in Contactors with One and Two Immobilized

524

Figure 23.7 Number of contactors Liqui Cel 400 2800 in MBSE and MBSS of phenylalanine, as well as the total number of contactors vs. yield of phenylalanine in MBSE [127]. The concentration ratio in the solvent at the raffinate end of the extractor was 0.60 (60%).

number. These results document the potential for application of HF contactors in recovery of organic acids from waste or fermentation solutions. Long-term evaluation of the MBSE/MBSS process for removal of chromium from ground water is presented in refs. [120, 121]. For ground water containing 774 g m3 of Cr(VI) the process showed stable performance during 700 h of operation of concentrating metal up to 20 kg m3 [120]. An integrated process for simultaneous removal of chromium from groundwater by MBSE followed by ion exchange is

Figure 23.8 Number of contactors Liqui Cel 400 2800 in MBSE and MBSS of phenylalanine, as well as the overall number of contactors vs. concentration ratio (approach to equilibrium) in the solvent at the raffinate end of the extractor [127]. Yield in MBSE was 70%.

23.4 Pertraction through BLME

suggested in ref. [121]. Laboratory and pilot-plant studies were performed with HF contactors with surface areas of ﬁbers of 1.4 and 19.3 m2, respectively. The design of the network of HF contactors and sensitivity analysis is presented in ref. [128]. Integration of the MBSE process with an ion-exchange step using Lewatit MP64 resin allowed Cr concentrations below 0.5 g m3 [121] to be achieved. Comparison of techniques for removal of chromium by MBSE, ion exchange and pertraction through BLME was carried out by Galan et al. [119].

23.4 Pertraction through BLME

In pertraction through BLME (Figure 23.1d) dispersion or emulsion of stripping solution in bulk of liquid membrane is used. In this way, compared to classical BLM (Figure 23.1c), one microporous wall is avoided. On the other hand, the advantage of the nondispersive process is lost. In systems inclined to emulsiﬁcation in stripping this may introduce a problem. Mass-transfer resistance will be lower in BLME than in the BLM system in Figure 23.1c. The ﬁrst papers on removal of copper from water by pertraction through BLM into a stripping solution emulsion were [129, 130]. A review of pertraction through BLME is given in refs. [131, 132]. PT through BLME is advantageous compared to PT through BLM or combination of MBSE with MBSS [119, 133]. Emulgation of stripping solution could be a problem in systems sensitive to formation of stable emulsions, but it is not clear how this phenomenon can inﬂuence this process. First, the nondispersive character of processes with immobilized interfaces is usually declared as an advantage. A lower loss of organic phase is connected with this. In most papers emulsion is not stabilized by a surfactant. In some papers [111, 134] a surfactant 3% v/v of Pluronic PE 3100 is used to enhance the phase separation of the aqueous stripping solution from the organic solution. By contrast, Fuad [135] added to the membrane phase emulsiﬁer Span 80 to stabilize emulsion and increasing the concentration of surfactant improved the zinc removal rate. This observation could be related to decreased droplet size in the emulsion and their increased surface area in emulsion. This could play a role in systems with slower kinetics of the stripping reaction. Generally, the presence of surfactant decrease the mass-transfer rate through a L/L interface due to resistance of the surfactant adsorption ﬁlm as shown, for example, in refs. [136, 137]. Modeling and optimization of pertraction into emulsion in HF contactors is discussed in refs. [77, 138]. The design and optimization of a network of HF contactors with minimum cost that permits the selective separation and recovery of anionic pollutants, for example, Cr(VI), using BLME process for groundwater remediation is presented in ref. [139] and for waste-water treatment in ref. [140]. Comparison of PT through BLME and MBSE with MBSS both in CF HFmodules is given in ref. [138]. Galan et al. [119] compared techniques for removal of chromium by MBSE with ion exchange and pertraction through BLME. An overview of selected papers on pertraction through BLME is presented in Table 23.3.

j525

j 23 Extractive Separations in Contactors with One and Two Immobilized

526

Table 23.3 Selected papers on recovery or removal of metals and

organics by pertraction through BLME.

Solute Cu(II)

Co(II) Cr(VI)

Cr(III) Zn(II)

Cd(II) Cu(II), Ni(II), Hg(I) Ni(II) Pu(IV) Sr(II) Au(I) Phenols Penicilin G Fumaric acid

Membrane phasea (extractant/ diluent) LIX 63/kerosene LIX 973N/(dodecane, dodecanol) LIX622N/kerosene LIX 64/kerosene Cyanex 301/(Isopar, dodecanol) Amberlite LA-2/(dodecane, dodecanol) Aliquat 336/(kerosene, isodecanol) Alamine 336/(Isopar L, dodecanol, surfactant) TOMAC/(n-decane, n-decanol) Cyanex 301/(dodecane, dodecanol) TBP (pure) n.a., full-scale units D2EHPA/isododecane (D2EHPA or trialkylphosphineoxide)/kerosene (LIX 84, D2EHPA, oleic acid)/tetradecane (without and with surfactant) di(2-butyloctyl)monothiophosphoric acid/(dodecane, dodecanol) TBP/dodecane (D2EHPA, 2-butyloctyl phenylphosphonic acid)/dodecane LIX79/n-heptane kerosene with surfactant Amberlite LA-2/Isopar Trialkylamine N7301/(kerosene with octanol)

Contactor typea

Literature

PF HF CF HF CF HF PF HF CF HF CF HF

[129, 130] [141–143] [133, 144] [145] [143] [143, 146]

CF HF CF HF

[119] [111, 134]

CF HF CF HF CF HF CF HF CF HF ﬂat membrane

[147] [141–143] [138] [37, 100] [135] [148]

PF HF

[149]

CF HF

[141]

HF CF HF

[150] [143, 151]

CF HF PF HF CF HF PF HF

[152, 153] [154] [143] [155]

a

Abbreviations used are explained in the list of abbreviations.

23.4.1 Case Studies

Selective removal of Zn and Fe from the passivating bath in the galvanic industry is important for preventing their accumulation in bath, and the need for bath replacement [37, 100]. Two full-scale installations of pertraction through BLME in galvanic companies proved the feasibility of this application in prolonged operation with an estimated payback time of less than 2 years [100]. Carrera et al. [138] reports the kinetics of zinc recovery from spent pickling solutions by pertraction through BLME in HF contactor with crossﬂow of phases. Pure tributyl phosphate was used as solvent and water as the stripping solution. Mass-transfer characteristics of the contactor have been estimated.

23.5 Pertraction through SLM Table 23.4 Selected papers on recovery or removal of metals and

organics by pertraction through SLM. Solute

Membrane phasea (extractant/diluent)

Contactor typea Literature

Cr(VI) Cu(II)

TBP/kerosene LIX54/kerosene D2EHPA/kerosene; SLM renewal technique Cyanex 302/kerosene D2EHPA/kerosene Cyanex 272/(kerosene and TBP as modiﬁer) PC88A (2-ethylhexyl phosphonic acid mono 2-ethylhexyl ester) TBP/kerosene Aliquat 336/(kerosene, 2-ethyl-1-hexanol) D2EHPA/dichlormethane and TOPO/ethyl acetate (Aliquat 336 in carbonate form)/kerosene; Stripping solution: aqueous 1M Na2CO3 Cyphos IL-104/n-dodecane Peracylated cyclodextrin TA-b-CD/ chloroform

PF HF CF HF PF HF

[181] [186, 187] [182]

Flat sheet CF HF Flat sheet

[17, 188] [189] [190] [191]

CF HF PF HF Flat sheet

[192] [193, 194] [195]

Flat sheet

[196]

Flat sheet Flat sheet

[44, 45] [197]

Cd(II) Co(II) Ga(III), As(III) U(VI) Zr(IV), Hf(IV) Fumaric and L-malic acids Lactic acid

Propranolol isomers separation a

Abbreviations used are explained in the list of abbreviations.

Copper catalyst was recovered from waste water from the wet peroxide oxidation process by pertraction through BLME and copper sulfate from stripping solution can be recycled to the reactor without loss in performance [144]. Ho et al. [141–143, 146, 151] published a series of papers on removal/recovery of several metals from waste solutions by pertraction through BLME in HF contactors as shown in Table 23.4. For some metals, such as zinc and copper, scale-up of this system to pilot plant with a HF module with a surface area of ﬁbers of 19 m2 (with diameter 10.2 and length 71.1 cm) was done and mass-transfer characteristics have been estimated [142]. Separation of phases in the dispersion from the stripping was satisfactory.

23.5 Pertraction through SLM

Supported liquid membranes were one of the ﬁrst types of liquid membranes published in 1967 [1, 2]. There are several types of SLM, which have an origin in three basic approaches: a. LM is soaked in micropores of the support, which is more or less inert. b. Liquid swells the polymer and forms a plasticized or gelled ﬁlm with interpenetrating continuous LM and polymer phases. A membrane can be cast from

j527

j 23 Extractive Separations in Contactors with One and Two Immobilized

528

polymer solution containing liquid carrier and volatile solvent, which after evaporation of volatiles, forms SLM. This type of SLMs are termed plasticized membranes [1, 156]. The most frequent terms used nowadays are polymer inclusion membrane (PIM) [157–160] or activated composite membranes [161–164]. A recent review on PIM is given in ref. [165]. c. LM is between two nonporous polymeric ﬁlms [2] or ﬁlms with reversed wettability, for example, organic LM between hydrophilic ﬁlms [69, 166]. This type of SLM belongs more to BLM because of the thickness of the membrane. There are several intermediate forms of these types, for example, pores of the support ﬁlled with a gel of polymer with LM [167] or SLM is covered on a surface(s) by thin ﬁlm(s) [167–171]. The preferred geometry for application of SLM is a hollow-ﬁber contactor because of the high surface area per unit volume of contactor. Thousands of papers on separation with SLM document their potential, but there is one great problem in their larger-scale applications connected with their short lifetime. Their performance in terms of more or less sharply declining solute ﬂux do not allowed wider utilization of SLM in practice. There are several mechanisms of ﬂux decline, but most important is partial solubility of the carrier and other components of SLM in aqueous phases in contact with it and emulgation of organic membrane [172–177]. Even when the solubility of SLM components is only minute it destroys its performance due to the several orders of magnitude difference between the volume of SLM and liquids brought into contact with it. There are several techniques that have been attempted to maintain stable performance of SLM, but a deﬁnite solution of this puzzle is not available yet. Covering of one or both SLM surface by a protective ﬁlm was tested in refs. [167–171, 178, 179]. Modules with continuous regeneration of SLM have been proposed [180–182]. Nakano [180] suggested equipment in which the membrane phase creeps up the walls of a vertical hollow ﬁbers. This enables a regeneration of the membrane phase in the pores. In pertraction of cobalt, the membrane broke down after about 50 h, while using a continuous regeneration the ﬂow did not decrease even after 150 h of operation. An SLM renewal technique with dosing droplets of membrane phase into the feed ﬂowing in ﬁber lumen suggest Ren et al. [181, 182]. This approach introduces dispergation of organic phase into the feed, which can result in membrane phase loss and at the same time contamination of the rafﬁnate. Tailoring of hollow ﬁbers for SLM was discussed in refs. [171, 178, 179]. Phosphonium ionic liquids can be a reactive carrier of organic acids and form effective SLM, as was found recently [183]. SLM with ionic liquid trihexyl-(tetradecyl) phosphonium bis 2,4,4-trimethylpentylphosphinate (Cyphos IL-104) had stable performance in pertraction of lactic acid for 5.3 days [44, 45], which is promising. Pertraction through SLM is widely used in analytical chemistry for separation and preconcentration of solutes before application of selected analytical method and it is discussed in refs. [32–34, 184, 185]. An overview of selected papers on recovery/removal of metals and organic compounds by pertraction through SLM is presented in Table 23.4.

23.7 Outlook

23.5.1 Case Studies

A pilot plant on recovery of copper from ammoniacal etching solutions was run with a good performance, as described in refs. [186, 187]. LIX54 in kerosene was used as the membrane phase in SLM. Large pilot-plant HF contactors Liqui cel (diam. 25.4 cm, length 71.1 cm) with surface area of 130 m2 have also been used [186]. A simulation of the hybrid fermentation–pertraction process for production of butyric acid shows that the pH of fermentation and pertraction should be optimized independently [198]. It is advantageous to have the pH of the feed into pertraction at about 4.0 for both IL and TOA carriers. Choosing a proper carrier in the supported liquid membrane between IL and TOA should be made according to actual operation conditions, because of the different transport properties of these carriers in respect to the concentration of undisociated form of BA. While at lower BA concentrations the IL is better, at higher concentrations of above 20 kg m3 and pH equal to 4.0, the membrane area needed is lower for TOA. An important factor will be the toxicity of the carrier to biomass. TOA is not very good in this respect and data for IL used are not available, but it is hoped that IL will be less toxic.

23.6 Comparison of Extractive Processes in HF Contactors and Pertraction through ELM

The advantages and disadvantages of membrane based processes and pertraction through various types of liquid membranes are summarized in Table 23.5. HF contactors are supposed in these processes with the exception of pertraction into stable emulsions (ELM) where mixed column contactors or mixer-settlers are used.

23.7 Outlook

Separation processes with one or two immobilized L/L interfaces realized in contactors with microporous or gel supports did not ﬁnd wider application up to now. The present state of knowledge supports the belief that they have potential for development of successful applications. The properties of the support wall(s), despite the fact that they do not play an active role in separation itself, have great importance in the achievement of reliable lifetime of the system and should be tailored or modiﬁed for the respective extractant or carrier used in separation. Speciﬁc surface modiﬁcation could be of great importance in achieving long lifetimes of separation systems. New extractants or carriers, for example, ionic liquids, could give a new impulse in this area. There is a need for advanced contactors with better stability in systems with organic phase. Deeper understanding of phenomena connected with achievement of prolonged operation of multiphase system involved in separation can help in development of successful applications. Better documented experimental

j529

j 23 Extractive Separations in Contactors with One and Two Immobilized

530

Table 23.5 Advantages and disadvantages of membrane-based

processes and pertraction through various types of liquid membranes in two- and three-phase systems.

Process MBSE or MBSS

PT through BLM

PT through BLME

Number of phases

Advantages

Disadvantages

2

- Nondispersive process

- Resistance of immobilizing wall. Two wall in MBSE and MBSS circuit replacing function of SLM

3

3

PT through SLM

3

PT through ELM

3

- No problems with stabile performance of the solvent - One immobilizing wall - Volume ratio of phases can be varied without limitations - No problems with stabile performance of BLM - Volume ratio of phases can be varied without limitations - One immobilizing wall

- No problems with stabile performance of LM - Larger surface are of the stripping solution droplets - Volume ratio of phases can be varied without limitations One immobilizing wall for two immobilized interfaces - Volume ratio of phases can be varied without limitations - Very small volume of membrane phase - No immobilizing wall - Comparatively high ﬂuxes can be achieved - Small volume of membrane phase

- Two immobilizing walls - Commercial contactors are not available. - Dispergation of stripping solution can introduce a problem in some systems sensitive to emulgation

- Limited stability of SLM

- Limited stability of emulsion - Swelling of emulsion - Complexity of the process - Resistance of the adsorption ﬁlm of surfactant

data on the long-term operation of separation systems are needed to achieve this. Knowledge accumulated in several pilot-plant experiments with larger HF modules may help to increase further the number of applications. Development of hybrid production–separation processes may be the way to successful solutions especially in the case of higher value added compounds.

References

Extractive fermentations and biotransformations could be a good example of these. Also, combinations of chemical reaction(s) with separation could have a synergy effect for the production part of technology enhancing its rate and yield.

Abbreviations

(ATPS) BLM BLME CF HF D2EHPA DMCCA ELM HF Hostarex A327 LM MBSE MBSS (ML) MPCA n.a. (PAA) PF HF (Phe) PIM PT (RD) SLM TBP TOA TOMAC TOPO

aqueous two-phase system bulk liquid membrane bulk liquid membrane with emulsion of the stripping solution crossﬂow hollow-ﬁber contactor di-2-ethylhexylphosphoric acid dimethylcyclopropancarboxylic acid emulsion liquid membrane hollow ﬁber (contactor) mixture of trialkylamines liquid membrane membrane-based solvent extraction membrane-based solvent stripping mother liquor 5-methyl-2-pyrazinecarboxylic acid not available phenylacetic acid parallel ﬂow hollow-ﬁber contactor phenylalanine polymer inclusion membrane pertraction rotary disc supported liquid membrane tributylphosphate trioctylamine trioctylmethylammonium chloride tri-n-octyl phosphine oxide

Acknowledgement

Support of the Slovak grant VEGA No. 1/0876/08 is acknowledged.

References 1 Bloch, R., Finkelstein, A., Kedem, O. and Vofsi, D. (1967) Metal-ion separation by

dialysis through solvent membranes. Ind Eng Chem Proc Des Develop, 6, 231.

j531

j 23 Extractive Separations in Contactors with One and Two Immobilized

532

2 Ward, W.J.I. and Robb, W.L. (1967) Carbon dioxide-oxygen separation: Facilitated transport of carbon dioxide across a liquid ﬁlm. Science, 156, 1481. 3 Li, N.N. (1968) Separating Hydrocarbons with Liquid Membranes, U.S. pat. 3,410,794. 4 Kim, B.M. (1984) Membrane-based solvent extraction for selective removal and recovery of metals. Journal of Membrane Science, 21, 5. 5 Sengupta, A., Basu, R. and Sirkar, K.K. (1988) Separation of solutes from aqueous-solutions by contained liquid membranes. AICHE Journal, 34, 1698. 6 Majumdar, S. and Sirkar, K.K. (1992) Hollow - ﬁber contained liquid membrane, in Membrane Handbook (eds W.S.W. Ho and K.K. Sirkar), Van Nostrand Reinhold, New York, p. 764. 7 Schlosser, Š., Rothova, I. and Frianova, H. (1993) Hollow-ﬁber pertractor with bulk liquid membrane. Journal of Membrane Science, 80, 99. 8 Schlosser, Š. (2000) Pertraction through liquid and polymeric membranes, in Integration of Membrane Processes into Bioconversions (eds K. Bako, L. Gubicza and M. Mulder), Kluwer Academic, p. 73. 9 Schlosser, Š. and Sabolova, E. (2002) Three phase contactor with distributed U-shaped bundles of hollow ﬁbers for pertraction. Journal of Membrane Science, 210, 331. 10 Livingston, A.G., Arcangeli, J.P., Boam, T., Zhang, S., Marangon, M. and Freitas, L.M. (1998) Extractive membrane bioreactors for detoxiﬁcation of chemical industry wastes: process development. Journal of Membrane Science, 151, 29. 11 Han, S., Ferreira, F.C. and Livingston, A. (2001) Membrane aromatic recovery system (MARS) – a new membrane process for the recovery of phenols from wastewaters. Journal of Membrane Science, 188, 219. 12 Ferreira, F.C., Peeva, L., Boam, A., Zhang, S. and Livingston, A. (2005) Pilot scale application of the Membrane Aromatic

13

14

15

16

17

18

19

20

21 22

Recovery System (MARS) for recovery of phenol from resin production condensates. Journal of Membrane Science, 257, 120. Draxler, J. and Marr, R. (1986) Emulsion liquid membranes part I: Phenomenon and industrial application. Chemical Engineering and Processing, 20, 319. Marr, R.J. and Draxler, J. (1992) Emulsion liquid membranes: applications, in Membrane Handbook (eds W.S.W. Ho and K.K. Sirkar), Van Nostrand Reinhold, New York, p. 701. Ho, W.S.W. and Li, N.N. (1996) Recent advances in emulsion liquid membranes, in Chemical Separations with Liquid Membranes (eds R.A. Bartsch and J.D. Way), ACS Symp. Series, ACS, Washington, p. 208. Perera, J.M. and Stevens, G.W. (2009) Use of emulsion liquid membrane systems in chemical and biotechnological separations, in Handbook of Membrane Separations (eds A.K. Pabby, S.S.H. Rizvi and A.M. Sastre), CRC Press, p. 709. Urtiaga, A.M., Alonso, A., Ortiz, I., Daoud, J.A., Elreefy, S.A., Deortiz, S.P. and Gallego, T. (2000) Comparison of liquid membrane processes for the removal of cadmium from wet phosphoric-acid. Journal of Membrane Science, 164, 229. Schlosser, Š., Sabolova, E., Kertesz, R. and (2001) Factors inﬂuencing Kubišova, L. transport through liquid membranes and membrane-based solvent-extraction. Journal of Separation Science, 24, 509. Lo, T.H., Baird, M.H.I. and Hanson, C. (1983) Handbook of Solvent Extraction, J. Wiley. Araki, T., and Tsukube, H. (1990) Liquid Membranes: Chemical Applications, CRC Press, Boca Raton, Florida, p. 224. Schugerl, K. (1994) Solvent Extraction in Biotechnology, Springer - Verlag, Berlin. Rethwisch, D.G., Parida, S., Yi, G. and Dordick, J.S. (1994) Use of alcohols as cosolvents in enzyme-facilitated transport of organic-acids through a liquid

References

23

24

25

26

27

28

29

30

31

32

membrane. Journal of Membrane Science, 95, 83. López, J.L. and Matson, S.L. (1997) A multiphase/extractive enzyme membrane reactor for production of diltiazem chiral intermediate. Journal of Membrane Science, 125, 189. Dai, X.P., Yang, Z.F., Luo, R.G. and Sirkar, K.K. (2000) Lipase-facilitated separation of organic-acids in a hollow-ﬁber contained liquid membrane module. Journal of Membrane Science, 171, 183. Miyako, E., Maruyama, T., Kamiya, N. and Goto, M. (2003) Enzyme-facilitated enantioselective transport of (S)ibuprofen through a supported liquid membrane based on ionic liquids. Chemical Communications, 2926. Schugerl, K. (2000) Integrated processing of biotechnology products. Biotechnology Advances, 18, 581. Schlosser, Š. (2000) Membrane based processes with immobilized interface, in Integration of Membrane Processes into Bioconversions (eds K. Bako, L. Gubicza and M. Mulder), Kluwer Academic, p. 55. Malinowski, J.J. (2001) Two-phase partitioning bioreactors in fermentation technology. Biotechnology Advances, 19, 525. Fernandes, P., Prazeres, D.M.F. and Cabral, J.M.S. (2003) Membrane-assisted extractive bioconversions, in Advances in Biochemical Engineering/Biotechnology (eds), Springer, p. 115. Schlosser, S., Kertesz, R. and Martak, J. (2005) Recovery and separation of organic acids by membrane-based solvent extraction and pertraction - An overview with a case study on recovery of MPCA. Separation and Puriﬁcation Technology, 41, 237. Membrana, Liqui-Cel Membrane Contactors http://www.liquicel.com [approached 2008 January]. Jonsson, J.A. and Mathiasson, L. (2001) Membrane extraction in analyticalchemistry. Journal of Separation Science, 24, 495.

33 Jonsson, J.A. (2003) Membrane extraction for sample preparation - a practical guide. Chromatographia, 57, S317. 34 Belkhouche, N.E., Didi, M.A., Romero, R., Jonsson, J.A. and Villemin, D. (2006) Study of new organophosphorus derivates carriers on the selective recovery of M(II) and M(III) metals, using supported liquid membrane extraction. Journal of Membrane Science, 284, 398. 35 Schlosser, Š. (1997) Method and equipment for mass and heat transfer among several liquids (in Slovak), Slovak pat. No. 278547. 36 ter Meulen, B.P. (1991) Transfer Device for the Transfer of Matter and/or Heat from One Medium Flow to Another Medium Flow, WO 91/09668. 37 Klaassen, R. and Jansen, A.E. (2001) The membrane contactor: Environmental applications and possibilities. Environmental Progress, 20, 37. 38 Prasad, R. and Sirkar, K.K. (1992) Membrane-based solvent extraction, in Membrane Handbook (eds W.S.W. Ho and K.K. Sirkar), Van Nostrand Reinhold, New York, p. 727. 39 Reed, B.W., Semmens, M.J. and Cussler, E.L. (1995) Membrane contactors, in Membrane Separation Technology. Principles and Applications (eds R.D. Noble and S.A. Stern), Elsevier Science, Amsterdam, p. 467. 40 Gabelman, A. and Hwang, S.T. (1999) Hollow-ﬁber membrane contactors. Journal of Membrane Science, 159, 61. 41 Sirkar, K.K. (1996) Hollow ﬁber-contained liquid membranes for separations - an overview, in Chemical Separations with Liquid Membranes (eds R.A. Bartsch and J.D. Way), ACS, p. 222. 42 Tong, Y.P., Hirata, M., Takanashi, H. and Hano, T. (1999) Back-extraction of lacticacid with microporous hollow-ﬁber membrane. Journal of Membrane Science, 157, 189. 43 Cichy, W., Schlosser, S. and Szymanowski, J. (2001) Recovery of phenol with cyanex(R)-923 in membrane

j533

j 23 Extractive Separations in Contactors with One and Two Immobilized

534

44

45

46

47

48

49

50

51

52

53

extraction-stripping systems. Solvent Extraction and Ion Exchange, 19, 905. Martak, J. and Schlosser, Š. (2006) Pertraction of organic acids through liquid membranes containing ionic liquids. Desalination, 199, 518. Martak, J., Schlosser, Š. and Vlckova, S. (2007) Pertraction of lactic acid through supported liquid membranes containing phosphonium ionic liquid. Submited for publication in Journal of Membrane Science. Juang, R.S., Chen, J.D. and Huan, H.C. (2000) Dispersion-free membrane extraction - case-studies of metal-ion and organic-acid extraction. Journal of Membrane Science, 165, 59. Kertesz, R. and Schlosser, Š. (2005) Design and simulation of two phase hollow-ﬁber contactors for simultaneous membranebased solvent extraction and stripping of organic acids and bases. Separation and Puriﬁcation Technology, 41, 275. Bobok, D., Schlosser, Š. and Kossacsky, E. (1984) Method of Separation of Solutions and Related Equipment for Realisation of this Method (in Slovak), Czechoslov. pat. 235362. Schlosser, Š. and Kossaczky, E. (1986) Pertraction through liquid membranes. Journal of Radioanalytical and Nuclear Chemistry, 101, 115. Boyadzhiev, L. and Atanassova, I. (1991) Recovery of L-Lysine from dilute water solutions by liquid pertraction. Biotechnology and Bioengineering, 38, 1059. Boyadzhiev, L. and Atanassova, I. (1994) Extraction of phenylalanine from dilutesolutions by rotating ﬁlm pertraction. Process Biochemistry, 29, 237. Boyadzhiev, L. and Kirilova, N. (2000) Extraction of tylosin from its aqueoussolutions by rotating ﬁlm-pertraction. Bioprocess Engineering, 22, 373. Boyadzhiev, L. and Dimitrova, V. (2006) Extraction and liquid membrane preconcentration of rosmarinic acid from lemon balm (Melissa ofﬁcinalis L.). Separation Science and Technology, 41, 877.

54 Boyadzhiev, L., Alexandrova, S., Kirilova, N. and Saboni, A. (2003) Pertraction continue de tylosine dans un contacteur a ﬁlms tournants. Chemical Engineering Journal, 95, 137. 55 Boyadzhiev, L. and Yordanov, B. (2004) Pertraction of indole alkaloids from Vinca minor L. Separation Science and Technology, 39, 1321. 56 Dimitrov, K., Metcheva, D. and Boyadzhiev, L. (2005) Integrated processes of extraction and liquid membrane isolation of atropine from Atropa belladonna roots. Separation and Puriﬁcation Technology, 46, 41. 57 Dimitrov, K., Metcheva, D., Alexandrova, S. and Boyadzhiev, L. (2006) Selective recovery of tropane alkaloids applying liquid membrane technique. Chemical and Biochemical Engineering Quarterly, 20, 55. 58 Lazarova, M. and Dimitrov, K. (2009) Selective recovery of alkaloids from glaucium ﬂavum crantz using integrated process extraction-pertraction. Separation Science and Technology, 44, 227. 59 Dimitrov, K., Gancel, F., Montastruc, L. and Nikov, I. (2008) Liquid membrane extraction of bio-active amphiphilic substances: Recovery of surfactin. Biochemical Engineering Journal, 42, 248. 60 Boyadzhiev, L. and Alexandrova, S. (1994) Recovery of copper from ammoniacal solutions by rotating ﬁlm pertraction. Hydrometallurgy, 35, 109. 61 Boyadzhiev, L. and Dimitrov, K. (1996) Silver recovery by rotating ﬁlm pertraction - process modeling. Solvent Extraction and Ion Exchange, 14, 105. 62 Dimitrov, K., Alexandrova, S., Saboni, A., Boyadzhiev, L. and Debray, E. (2002) Separation of metals by rotating ﬁlm pertraction. Chemical Engineering & Technology, 25, 823. 63 Zhivkova, S., Dimitrov, K., Kyuchoukov, G. and Boyadzhiev, L. (2004) Separation of zinc and iron by pertraction in rotating ﬁlm contactor with Kelex 100 as a carrier. Separation and Puriﬁcation Technology, 37, 9.

References 64 Dimitrov, K., Rollet, V., Saboni, A. and Alexandrova, S. (2008) Recovery of nickel from sulphate media by batch pertraction in a rotating ﬁlm contactor using Cyanex 302 as a carrier. Chemical Engineering and Processing, 47, 1562. 65 Prasad, R. and Sirkar, K.K. (1988) Dispersion-free solvent extraction with microporous hollow-ﬁber modules. AICHE Journal, 34, 177. 66 Ding, H. and Cussler, E.L. (1991) Fractional extraction with hollow ﬁbers with hydrogel-ﬁlled walls. AICHE Journal, 37, 855. 67 Stark, D., Kornmann, H., Muench, T., Sonnleitner, B., Marison, I.W. and von Stockar, U. (2003) Novel type of in situ extraction: use of solvent containing microcapsules for the bioconversion of 2phenylethanol from L-Phenylalanine by Saccharomyces cerevisiae. Biotechnology and Bioengineering, 83, 376. 68 Kondo, K., Otono, T. and Matsumoto, M. (2004) Preparation of microcapsules containing extractants and the application of the microcapsules to the extractive fermentation of lactic acid. Journal of Chemical Engineering of Japan, 37, 1. 69 Kedem, O. and Bromberg, L. (1993) Ionexchange membranes in extraction processes. Journal of Membrane Science, 78, 255. 70 Isono, Y., Fukushima, K., Kawakatsu, T. and Nakajima, M. (1995) New selective perstraction system with charged membrane. Journal of Membrane Science, 105, 293. 71 Isono, Y., Fukushima, K., Kawakatsu, T. and Nakajima, M. (1997) Integration of charged membrane into perstraction system for separation of amino-acid derivatives. Biotechnology and Bioengineering, 56, 162. 72 Riedl, W. and Raiser, T. (2008) Membranesupported extraction of biomolecules with aqueous two-phase systems. Desalination, 224, 160. 73 Martak, J., Schlosser, Š., Sabolova, E., Krištofıkova, L. and Rosenberg, M. (2003)

74

75

76

77

78

79

80

81

Fermentation of lactic acid with Rhizopus arrhizus in a stirred tank reactor with a periodical bleed and feed operation. Process Biochemistry, 38, 1573. and Schlosser, Š. (1996) Kubišova, L. Pertraction and extraction of lactic acid in hollow-ﬁbre contactors. Proc. of 12th Int. Congr. Chisa 96, Praha, Czech Republic, full paper on CDROM. Sciubba, L., Di Gioia, D., Fava, F. and Gostoli, C. (2008) Membrane-based solvent extraction of vanilin in hollowﬁber contactors. accepted for publication in Desalination. Juang, R.S. and Huang, W.C. (2008) Use of membrane contactors as two-phase bioreactors for the removal of phenol in saline and acidic solutions. Journal of Membrane Science, 313, 207. Ortiz, I. and Irabien, A. (2009) Membrane-assisted solvent extraction for the recovery of metallic pollutants: process modeling and optimization, in Handbook of Membrane Separations (eds A.K. Pabby, S.S.H. Rizvi and A.M. Sastre), CRC Press, p. 1023. Reis, M.T.A., de Freitas, O.M.F., Ismael, M.R.C. and Carvalho, J.M.R. (2007) Recovery of phenol from aqueous solutions using liquid membranes with Cyanex 923. Journal of Membrane Science, 305, 313. Gameiro, M.L.F., Ismael, M.R.C., Reis, M.T.A. and Carvalho, J.M.R. (2008) Recovery of copper from ammoniacal medium using liquid membranes, with LIX 54. Separation and Puriﬁcation Technology, Correct. Proof. Escalante, H., Alonso, A.I., Ortiz, I. and Irabien, A. (1998) Separation of Lphenylalanine by nondispersive extraction and backextraction equilibrium and kinetic-parameters. Separation Science and Technology, 33, 119. Escalante, H. and Irabien, A. (2001) Separation/concentration of Lphenylalanine in hollow ﬁbre modules. decrease of the extraction rate, in Solvent Extraction for the 21st Century (eds M.

j535

j 23 Extractive Separations in Contactors with One and Two Immobilized

536

82

83

84

85

86

87

88

89

90

Valiente and M. Hidalgo), Society of Chemical Industries, London, p. 1499. Cardoso, M.M., Viegas, R.M.C. and Crespo, J.P.S.G. (1999) Extraction and reextraction of phenylalanine by cationic reversed micelles in hollow-ﬁber contactors. Journal of Membrane Science, 156, 303. Gerigk, M.R., Maass, D., Kreutzer, A., Sprenger, G., Bongaerts, J., Wubbolts, M. and Takors, R. (2002) Enhanced pilotscale fed-batch L-phenylalanine production with recombinant escherichia coli by fully integrated reactive extraction. Bioprocess and Biosystems Engineering, 25, 43. Maass, D., Gerigk, M.R., Kreutzer, A., Weuster-Botz, D., Wubbolts, M. and Takors, R. (2002) Integrated Lphenylalanine separation in an E-coli fedbatch process: from laboratory to pilot scale. Bioprocess and Biosystems Engineering, 25, 85. Takors, R. (2004) Model-based analysis and optimization of an ISPR approach using reactive extraction for pilot-scale L-phenylalanine production. Biotechnology Progress, 20, 57. Kertesz, R., Šimo, M. and Schlosser, Š. (2005) Membrane-based solvent extraction and stripping of phenylalanine in HF contactors. Journal of Membrane Science, 257, 37. Harington, T. and Hossain, M.M. (2008) Extraction of lactic acid into sunﬂower oil and its recovery into an aqueous solution. Desalination, 218, 287. Wu, Z.T. and Yang, S.T. (2003) Extractive fermentation for butyric acid production from glucose by clostridium tyrobutyricum. Biotechnology and Bioengineering, 82, 93. Lin, S.H., Chen, C.N. and Juang, R.S. (2006) Kinetic analysis on reactive extraction of aspartic acid from water in hollow-ﬁber membrane modules. Journal of Membrane Science, 281, 186. Rindﬂeisch, D., Syska, B., Lazarova, Z. and Schugerl, K. (1997) Integrated

91

92

93

94

95

96

97

98

membrane extraction, enzymic conversion and electrodialysis for the synthesis of ampicilin from penicilin G. Process Biochemistry, 32, 605. Lazarova, Z., Syska, B. and Schugerl, K. (2002) Application of large-scale hollowﬁber membrane contactors for simultaneous extractive removal and stripping of penicillin G. Journal of Membrane Science, 202, 151. Hossain, M.M. and Dean, J. (2008) Extraction of penicillin G from aqueous solutions: Analysis of reaction equilibrium and mass transfer. Separation and Puriﬁcation Technology, 62, 437. Ratajczak, W., Porebski, T., Plesnar, M., Brzozowski, R., Tomzik, S., Capala, W., Wieteska, A., Ebrowski, M., Malachowski, B. and Karabin, M. (2004) Modernization of the phenol and acetone plant at the Orlen Polish oil company. Przemysl Chemiczny, 83, 498. Ratajczak, W., Porebski, T., Plesnar, M. and Tomzik, S. (2007) Process backﬁtting in the Polish chemical industry. Modern unit operations. Przemysl Chemiczny, 86, 262. Kubišova, L., Sabolova, E., Schlosser, S., Martak, J. and Kertesz, R. (2002) Membrane-based solvent extraction and stripping of a heterocyclic carboxylic acid in hollow-ﬁber contactors. Desalination, 148, 205. Kubišova, L., Sabolova, E., Schlosser, Š., Martak, J. and Kertesz, R. (2004) Masstransfer in membrane-based solvent extraction and stripping of 5-methyl-2pyrazinecarboxylic acid and co-transport of sulphuric acid in HF contactors. Desalination, 163, 27. Schlosser, Š., Sabolova, E. and Martak, J. (2001) Pertraction and membrane-based solvent extraction of carboxylic acids in hollow-ﬁber contactors, in Solvent Extraction for the 21st Century (eds M. Valiente and M. Hidalgo), Society of Chemical Industries, London, p. 1041. Schlosser, Š. (2002) Method of separation of carboxylic acids from aqueous and

References

99

100

101

102

103

104

105

106

107

108

nonaqueous solutions (in Slovak), Slovak pat. No. 282775. Yang, X., Cao, Y.M., Wang, R. and Yuan, Q. (2007) Study on highly hydrophilic cellulose hollow-ﬁber membrane contactors for thiol sulfur removal. Journal of Membrane Science, 305, 247. Klaassen, R., Feron, P. and Jansen, A. (2008) Membrane contactors applications. Desalination, 224, 81. Baudot, A., Floury, J. and Smorenburg, H.E. (2001) Liquid-liquid extraction of aroma compounds with hollow-ﬁber contactor. AICHE Journal, 47, 1780. Pierre, F.X., Souchon, I. and Marin, M. (2001) Recovery of sulfur aroma compounds using membrane-based solvent-extraction. Journal of Membrane Science, 187, 239. Souchon, I., Athes, V., Pierre, F.X. and Marin, M. (2004) Liquid-liquid extraction and air stripping in membrane contactor: application to aroma compounds recovery. Desalination, 163, 39. Bocquet, S., Viladomat, E.G., Nova, C.M., Sanchez, J., Athes, V. and Souchon, I. (2006) Membrane-based solvent extraction of aroma compounds: Choice of conﬁgurations of hollow ﬁber modules based on experiments and simulation. Journal of Membrane Science, 281, 358. Younas, M., Bocquet, S.D. and Sanchez, J. (2008) Extraction of aroma compounds in a HFMC: Dynamic modelling and simulation. Journal of Membrane Science, 323, 386. Brose, D.J., Chidlaw, M.B., Friesen, D.T., Lachapelle, E.D. and Vaneikeren, P. (1995) Fractionation of citrus oils using a membrane-based extraction process. Biotechnology Progress, 11, 214. de Haan, A.B., Bartels, P.V. and de Graauw, J. (1989) Extraction of metal ions from waste water. Modelling of the mass transfer in a supported-liquid-membrane process. Journal of Membrane Science, 45, 281. Lin, S.H. and Juang, R.S. (2002) Kinetic modeling of simultaneous recovery of

109

110

111

112

113

114

115

116

117

metallic cations and anions with a mixture of extractants in hollow-ﬁber modules. Industrial & Engineering Chemistry Research, 41, 853. Koopman, C. and Witkamp, G.J. (2002) Extraction of heavy metals from industrial phosphoric acid in a transverse ﬂow hollow-ﬁber membrane contactor. Separation Science and Technology, 37, 1273. Ortiz, I., Bringas, E., San Roman, M.F. and Urtiaga, A.M. (2004) Selective separation of zinc and iron from spent pickling solutions by membrane-based solvent extraction: Process viability. Separation Science and Technology, 39, 2441. Ortiz, I., Bringas, E., Samaniego, H., San Roman, F. and Urtiaga, A. (2006) Membrane processes for the efﬁcient recovery of anionic pollutants. Desalination, 193, 375. Samaniego, H., San Roman, M.F. and Ortiz, I. (2007) Kinetics of zinc recovery from spent pickling efﬂuents. Industrial & Engineering Chemistry Research, 46, 907. Vajda, M., Kosuthova, A. and Schlosser, S. (2004) Membrane-based extraction joined with membrane-based stripping in a circulating arrangement - III. Extraction of zinc. Chemical Papers, 58, 1. Fouad, E.A. and Bart, H.J. (2007) Separation of zinc by a non-dispersion solvent extraction process in a hollowﬁber contactor. Solvent Extraction and Ion Exchange, 25, 857. Alonso, A.I., Urtiaga, A.M., Zamacona, S., Irabien, A. and Ortiz, I. (1997) Kinetic modelling of cadmium removal from phosphoric acid by non-dispersive solvent extraction. Journal of Membrane Science, 130, 193. Kao, H.C. and Juang, R.S. (2003) Hindered membrane diffusion in the nondispersive stripping of Co(II) from organic amine solutions with hydrochloric acid. Industrial & Engineering Chemistry Research, 42, 6181. Ortiz, I., Galan, B. and Irabien, A. (1996) Membrane mass-transport coefﬁcient for

j537

j 23 Extractive Separations in Contactors with One and Two Immobilized

538

118

119

120

121

122

123

124

125

the recovery of Cr(VI) in hollow-ﬁber extraction and back-extraction modules. Journal of Membrane Science, 118, 213. Alonso, A.I., Galan, B., Gonzalez, M. and Ortiz, I. (1999) Experimental and theoretical analysis of a nondispersive solvent extraction pilot plant for the removal of Cr(VI) from a galvanic process wastewaters. Industrial & Engineering Chemistry Research, 38, 1666. Galan, B., Castaneda, D. and Ortiz, I. (2005) Removal and recovery of Cr(VI) from polluted ground waters: A comparative study of ion-exchange technologies. Water Research, 39, 4317. Galan, B., Calzada, M. and Ortiz, I. (2006) Recycling of Cr(VI) by membrane solvent extraction: Long term performance with the mathematical model. Chemical Engineering Journal, 124, 71. Galan, B., Castaneda, D. and Ortiz, I. (2008) Integration of ion exchange and non-dispersive solvent extraction processes for the separation and concentration of Cr(VI) from ground waters. Journal of Hazardous Materials, 152, 795. Kumar, A., Haddad, R., Alguacil, F.J. and Sastre, A.M. (2005) Comparative performance of non-dispersive solvent extraction using a single module and the integrated membrane process with two hollow-ﬁber contactors. Journal of Membrane Science, 248, 1. Kumar, A., Haddad, R. and Sastre, A.M. (2001) Integrated membrane process for gold recovery from hydrometallurgical solutions. AICHE Journal, 47, 328. Patil, C.B., Mohapatra, P.K. and Manchanda, V.K. (2008) Transport of thorium from nitric acid solution by nondispersive solvent extraction using a hollow ﬁbre contactor. Desalination, 232, 272. Trtic, T.M., Vladisavljevic, G.T. and Comor, J.J. (2000) Dispersion-free solvent-extraction of thallium(III) in hollow-ﬁber contactors. Separation Science and Technology, 35, 1587.

126 Trtic, T.M., Vladisavljevic, G.T. and Comor, J.J. (2001) Single-stage and 2-stage solvent-extraction of Tl(III) in hollow-ﬁber contactors under recirculation mode of operation. Separation Science and Technology, 36, 295. 127 Kertesz, R. and Schlosser, Š. (2004) Simulation of simultaneous membranebased solvent extraction and stripping of phenylalanine in hollow-ﬁber contactors. Proc. 16th Int. Congr. CHISA 2004, Prague, full text on CD ROM. 128 Alonso, A.I. and Gruhn, G. (2002) Flexibility analysis of nondispersive solvent extraction plant. Separation Science and Technology, 37, 161. 129 Schneider, K. and Rintelen, T.H. (1986) Recovery of metals with the aid of supported liquid membranes. Chem Ing Tech, 58, 800. 130 Schulz, G. (1988) Separation techniques with supported liquid membranes. Desalination, 68, 191. 131 Pabby, A.K. and Sastre, A.M. (2006) Hollow-ﬁber membrane based separation technology: performance and design perspectives, in Solvent Extraction and Liquid Membranes: Fundamental and Application in New Materials (eds Cortina and Aguilar), Marcel Dekker, New York. 132 Zou, J., Huang, J. and Ho, W.S.W. (2008) Facilitated transport membranes for environmental energy, and biochemical applications, in Advanced Membrane Technology and Applications (eds N.N. Li, A.C. Fane, W.S.W. Ho and T. Miltsuura), Wiley, p. 721. 133 Urtiaga, A., Abellan, M.J., Irabien, J.A. and Ortiz, I. (2005) Membrane contactors for the recovery of metallic compounds: Modelling of copper recovery from WPO processes. Journal of Membrane Science, 257, 161. 134 Bringas, E., San Roman, M.F. and Ortiz, I. (2006) Separation and recovery of anionic pollutants by the emulsion pertraction technology. Remediation of polluted groundwaters with Cr(VI). Industrial & Engineering Chemistry Research, 45, 4295.

References 135 Fouad, E.A. and Bart, H. (2008) Emulsion liquid membrane extraction of zinc by a hollow-ﬁber contactor. Journal of Membrane Science, 307, 156. 136 Schlosser, Š. and Kossaczky, E. (1987) Liquid membranes stabilized by polymeric surfactants, in Synthetic Polymeric Membranes (ed. B. Sedlacek), W. de Gruyter, Berlin, p. 571. 137 Strzelbicki, J. and Schlosser, Š. (1989) Inﬂuence of surface-active substances on pertraction of cobalt(II) cations through bulk and emulsion liquid membranes. Hydrometallurgy, 23, 67. 138 Carrera, J.A., Bringas, E., Roman, M.F.S. and Ortiz, I. (2009) Selective membrane alternative to the recovery of zinc from hot-dip galvanizing efﬂuents. Journal of Membrane Science, 326, 672. 139 Bringas, E., Karuppiah, R., San Roman, M.F., Ortiz, I. and Grossmann, I.E. (2007) Optimal groundwater remediation network design using selective membranes. Industrial & Engineering Chemistry Research, 46, 5555. 140 San Roman, M.F., Bringas, E., Ortiz, I. and Grossmann, I.E. (2007) Optimal synthesis of an emulsion pertraction process for the removal of pollutant anions in industrial wastewater systems. Comp Chem Eng, 31, 456. 141 Ho, W.S.W., Wang, B., Neumuller, T.E. and Roller, J. (2001) Supported liquid membranes for removal and recovery of metals from waste waters and process streams. Environmental Progress, 20, 117. 142 Ho, W.S.W., Poddar, T.K. and Neumuller, T.E. (2002) Removal and recovery of copper and zinc by supported liquid membranes with strip dispersion. Journal of the Chinese Institute of Chemical Engineers, 33, 67. 143 Ho, W.S.W. (2003) Removal and recovery of metals and other materials by supported liquid membranes with strip dispersion. Annals of the New York Academy of Sciences, Advanced Membrane Technology, 97. 144 Urtiaga, A., Abellan, M.J., Irabien, A. and Ortiz, I. (2006) Use of membrane

145

146

147

148

149

150

151

152

contactors as an efﬁcient alternative to reduce efﬂuent ecotoxicity. Desalination, 191, 79. Hu, S.Y.B. and Wiencek, J.M. (1998) Emulsion-liquid-membrane extraction of copper using a hollow-ﬁber contactor. AICHE Journal, 44, 570. Ho, W.S.W. and Poddar, T.K. (2001) New membrane technology for removal and recovery of chromium from waste waters. Environmental Progress, 20, 44. Alguacil, F.J., Alonso, M., Lopez, F.A. and Lopez-Delgado, A. (2009) Application of pseudo-emulsion based hollow ﬁber strip dispersion (PEHFSD) for recovery of Cr (III) from alkaline solutions. Separation and Puriﬁcation Technology, In Press. Corrected Proof. He, D., Gu, S. and Ma, M. (2007) Simultaneous removal and recovery of cadmium (II) and CN- from simulated electroplating rinse wastewater by a strip dispersion hybrid liquid membrane (SDHLM) containing double carrier. Journal of Membrane Science, 305, 36. Raghuraman, B. and Wiencek, J. (1993) Extraction with emulsion liquid membranes in a hollow-ﬁber contactor. AICHE Journal, 39, 1885. Gupta, S.K., Rathore, N.S., Sonawane, J.V., Pabby, A.K., Venugopalan, A.K., Changrani, R.D., Dey, P.K. and Venkataramani, B. (2005) Application of hollow-ﬁber contactor in nondispersive solvent extraction of Pu(IV) by TBP. Separation Science and Technology, 40, 1911. Ho, W.S.W. and Wang, B. (2002) Strontium removal by new alkyl phenylphosphonic acids in supported liquid membranes with strip dispersion. Industrial & Engineering Chemistry Research, 41, 381. Sonawane, J.V., Pabby, A.K. and Sastre, A.M. (2007) Au(I) extraction by LIX-79/ n-heptane using the pseudo-emulsionbased hollow-ﬁber strip dispersion (PEHFSD) technique. Journal of Membrane Science, 300, 147.

j539

j 23 Extractive Separations in Contactors with One and Two Immobilized

540

153 Sonawane, J.V., Pabby, A.K. and Sastre, A.M. (2008) Pseudo-emulsion based hollow ﬁber strip dispersion: A novel methodology for gold recovery. AICHE Journal, 54, 453. 154 Nanoti, A., Ganguly, S.K., Goswami, A.N. and Rawat, B.S. (1997) Removal of phenols from wastewater using liquid membranes in a microporous hollow ﬁber-membrane extractor. Industrial & Engineering Chemistry Research, 36, 4369. 155 Li, S.-J., Chen, H.-L. and Zhang, L. (2009) Recovery of fumaric acid by hollow-ﬁber supported liquid membrane with strip dispersion using N7301 as carrier. Separation and Puriﬁcation Technology, doi:10.1016/j.seppur.2008.12.004. 156 Paugam, M.F. and Bufﬂe, J. (1998) Comparison of carrier-facilitated copper (II) ion transport mechanisms in a supported liquid membrane and in a plasticized cellulose triacetate membrane. Journal of Membrane Science, 147, 207. 157 Schow, A.J., Peterson, R.T. and Lamb, J.D. (1996) Polymer inclusion membranes containing macrocyclic carriers for use in cation separations. Journal of Membrane Science, 111, 291. 158 Gardner, J.S., Walker, J.O. and Lamb, J.D. (2004) Permeability and durability effects of cellulose polymer variation in polymer inclusion membranes. Journal of Membrane Science, 229, 87. 159 Kozlowski, C.A., Kozlowska, J., Pellowski, W. and Walkowiak, W. (2006) Separation of cobalt-60, strontium-90, and cesium-137 radioisotopes by competitive transport across polymer inclusion membranes with organophosphorous acids. Desalination, 198, 141. 160 Fontas, C., Tayeb, R., Dhahbi, M., Gaudichet, E., Thominette, F., Roy, P., Steenkeste, K., Fontaine-Aupart, M.P., Tingry, S., Tronel-Peyroz, E. and Seta, P. (2007) Polymer inclusion membranes: The concept of ﬁxed sites membrane revised. Journal of Membrane Science, 290, 62. 161 Oleinikova, M., Gonzalez, C., Valiente, M. and Munoz, M. (1999) Selective transport

162

163

164

165

166

167

168

169

of zinc through activated composite membranes containing di(2-ethylhexyl) dithiophosphoric acid as a carrier. Polyhedron, 18, 3353. Gumi, T., Oleinikova, M., Palet, C., Valiente, M. and Munoz, M. (2000) Facilitated transport of lead(II) and cadmium(II) through novel activated composite membranes containing di(2-ethyl-hexyl)phosphoric acid as carrier. Analytica Chimica Acta, 408, 65. Macanas, J. and Munoz, M. (2005) Mass transfer determining parameter in facilitated transport through di-(2ethylhexyl) dithiophosphoric acid activated composite membranes. Analytica Chimica Acta, 534, 101. Kozlowski, C.A. (2006) Facilitated transport of metal ions through composite and polymer inclusion membranes. Desalination, 198, 132. Nghiem, L.D., Mornane, P., Potter, I.D., Perera, J.M., Cattrall, R.W. and Kolev, S.D. (2006) Extraction and transport of metal ions and small organic compounds using polymer inclusion membranes (PIMs). Journal of Membrane Science, 281, 7. Wodzki, R. and Nowaczyk, J. (2002) Propionic and acetic acid pertraction through a multimembrane hybrid system containing TOPO or TBP. Separation and Puriﬁcation Technology, 26, 207. Kemperman, A.J.B., Damink, B., Vandenboomgaard, T. and Strathmann, H. (1997) Stabilization of supported liquid membranes by gelation with PVC. Journal of Applied Polymer Science, 65, 1205. Kemperman, A.J.B., Rolevink, H.H.M., Bargeman, D., Vandenboomgaard, T. and Strathmann, H. (1998) Stabilization of supported liquid membranes by interfacial polymerization top layers. Journal of Membrane Science, 138, 43. Wijers, M.C., Jin, M., Wessling, M. and Strathmann, H. (1998) Supported liquid membranes modiﬁcation with sulfonated poly(ether ether ketone) - permeability, selectivity and stability. Journal of Membrane Science, 147, 117.

References 170 Wang, Y.C. and Doyle, F.M. (1999) Formation of epoxy skin layers on the surface of supported liquid membranes containing polyamines. Journal of Membrane Science, 159, 167. 171 He, T. (2008) Towards stabilization of supported liquid membranes: preparation and characterization of polysulfone support and sulfonated poly (ether ether ketone) coated composite hollow-ﬁber membranes. Desalination, 225, 82. 172 Danesi, P.R., Reichley-Yinger, L. and Rickert, P.G. (1987) Lifetime of supported liquid membranes: The inﬂuence of interfacial properties, chemical composition and water transport on the long term stability of the membranes. Journal of Membrane Science, 31, 117. 173 Neplenbroek, A.M., Bargeman, D. and Smolders, C.A. (1992) Supported liquid membranes – instability effects. Journal of Membrane Science, 67, 121. 174 Neplenbroek, A.M., Bargeman, D. and Smolders, C.A. (1992) Mechanism of supported liquid membrane degradation – emulsion formation. Journal of Membrane Science, 67, 133. 175 Kemperman, A.J.B., Bargeman, D., Vandenboomgaard, T. and Strathmann, H. (1996) Stability of supported liquid membranes - state-of-the-art. Separation Science and Technology, 31, 2733. 176 Dreher, T.M. and Stevens, G.W. (1998) Instability mechanisms of supported liquid membranes. Separation Science and Technology, 33, 835. 177 Zhang, B., Gozzelino, G. and Baldi, G. (2001) Membrane liquid loss of supported liquid membrane-based on N-decanol. Colloids and Surfaces A, 193, 61. 178 He, T., Versteeg, L.A.M., Mulder, M.H.V. and Wessling, M. (2004) Composite hollow-ﬁber membranes for organic solvent-based liquid-liquid extraction. Journal of Membrane Science, 234, 1. 179 Yang, Q., Chung, T.S., Xiao, Y.C. and Wang, K.Y. (2007) The development of chemically modiﬁed P84Co-polyimide membranes as supported liquid

180

181

182

183

184

185

186

187

membrane matrix for Cu(II) removal with prolonged stability. Chemical Engineering Science, 62, 1721. Nakano, M., Takahashi, K. and Takeuchi, H. (1987) A method for continuous operation of supported liquid membranes. Journal of Chemical Engineering of Japan, 20, 326. Ren, Z.Q., Liu, J.T., Zhang, W.D., Du, C.S. and Ma, J.N. (2006) Treatment of wastewater containing chromiun (VI) by hollow ﬁber renewal liquid membrane technology. Electroplating and Finishing, 25, 49. Ren, Z.Q., Zhang, W.D., Liu, Y.M., Dai, Y. and Cui, C.H. (2007) New liquid membrane technology for simultaneous extraction and stripping of copper(II) from wastewater. Chemical Engineering Science, 62, 6090. Martak, J. and Schlosser, Š. (2007) Extraction of lactic acid by phosphonium ionic liquids. Separation and Puriﬁcation Technology, 57, 483. Chimuka, L., Cukrowska, E. and Jonsson, J.A. (2004) Why liquid membrane extraction is an attractive alternative in sample preparation. Pure and Applied Chemistry, 76, 707. Bardstu, K.F., Ho, T.S., Rasmussen, K.E., Pedersen-Bjergaard, S. and Jonsson, J.A. (2007) Supported liquid membranes in hollow ﬁber liquid-phase microextraction (LPME) – Practical considerations in the three-phase mode. Journal of Separation Science, 30, 1364. Yang, Q. and Kocherginsky, N.M. (2006) Copper recovery and spent ammoniacal etchant regeneration based on hollow ﬁber supported liquid membrane technology: From bench-scale to pilotscale tests. Journal of Membrane Science, 286, 301. Yang, Q. and Kocherginsky, N.M. (2007) Copper removal from ammoniacal wastewater through a hollow ﬁber supported liquid membrane system: Modeling and experimental veriﬁcation. Journal of Membrane Science, 297, 121.

j541

j 23 Extractive Separations in Contactors with One and Two Immobilized

542

188 Daoud, J.A., Elreefy, S.A. and Aly, H.F. (1998) Permeation of Cd(II) ions through a supported liquid membrane containing Cyanex-302 in kerosene. Separation Science and Technology, 33, 537. 189 Prakorn, R., Eakkapit, S., Weerawat, P., Milan, H. and Pancharoen, U. (2006) Permeation study on the hollow-ﬁber supported liquid membrane for the extraction of cobalt(II). Korean Journal of Chemical Engineering, 23, 117. 190 Swain, B., Jeong, J., Lee, J.C. and Lee, G.H. (2007) Extraction of Co(II) by supported liquid membrane and solvent extraction using Cyanex 272 as an extractant: A comparison study. Journal of Membrane Science, 288, 139. 191 Tsai, C.Y., Chen, Y.F., Chen, W.C., Yang, F.R., Chen, J.H. and Lin, J.C. (2005) Separation of gallium and arsenic in wafer grinding extraction solution using a supported liquid membrane that contains PC88A as a carrier. Journal of Environmental Science and Health Part A-Toxic/Hazardous Substances & Environmental Engineering, 40, 477. 192 Ramakul, P., Prapasawad, T., Pancharoen, U. and Pattaveekongka, W. (2007) Separation of radioactive metal ions by hollow ﬁber-supported liquid membrane and permeability analysis. Journal of the Chinese Institute of Chemical Engineers, 38, 489.

193 Yang, X.J., Fane, A.G. and Pin, C. (2002) Separation of zirconium and hafnium using hollow ﬁbers - Part I. Supported liquid membranes. Chemical Engineering Journal, 88, 37. 194 Yang, X.J., Fane, A.G. and Soldenhoff, K. (2003) Comparison of liquid membrane processes for metal separations: Permeability, stability, and selectivity. Industrial & Engineering Chemistry Research, 42, 392. 195 Bressler, E., Pines, O., Goldberg, I. and Braun, S. (2002) Conversion of fumaric acid to L-malic by sol-gel immobilized Saccharomyces cerevisiae in a supported liquid membrane bioreactor. Biotechnology Progress, 18, 445. 196 Yang, Q. and Chung, T.S. (2007) Modiﬁcation of the commercial carrier in supported liquid membrane system to enhance lactic acid ﬂux and to separate L, D-lactic acid enantiomers. Journal of Membrane Science, 294, 127. 197 Ferreira, Q., Coelhoso, I.M., Ramalhete, N. and Marques, H.M.C. (2006) Resolution of racemic propranolol in liquid membranes containing TA-betacyclodextrin. Separation Science and Technology, 41, 3553. 198 Blahušiak, M., Schlosser, Š. and Martak, J. (2009) Simulation of hybrid fermentationseparation process for production of butyric acid. Submitted for publication in Chemical Papers.

j543

Index a acid average concentrations 112 activated sludge 363 additives 30 adhesion 434 adsorption/adhesion 33 aeration 377, 380 aging permeability change 71 aging phenomenon 73 aging rate 71, 73 air dehumidiﬁcation 502, 505 air drying 173 amino acids 399 amorphous polymer 4, 23, 64, 68 amorphous silica membranes 208 anchorage-dependent cells 434 anion-exchange membranes 84 anisotropic membrane 21 anisotropic polymer 28 aroma recovery 253 aromatics 182 artiﬁcial liver 426 artiﬁcial organ 411, 426 asymmetric anisotropic membrane 21 asymmetric membranes 63 asymmetric skin layer 69 atomic radius 5 atomic weight 5 atomistic molecular modeling 13 atomistic packing models 8 average resistance 104

b backbone polymers 38 backwashing 383 Baxter autopheresis C 421 best available technology 367, 386

bioactive peptides recovery 254 bioartiﬁcial liver 429 bioartiﬁcial organ 411, 426 biocatalysts 397 biocatalytic membrane reactors 402 biocatalytic reactions 315 biochemical membrane reactors 309, 397 bioethanol 189 bioﬁlms 128 biofouling 128, 130, 132, 236 biohybrid system 435 biological operating conditions 379 biological origin catalyst 285 biological treatment 379 biomass 379 biomaterials 441 biomedical materials 433 biomedical membrane 411 bioreactors 438 biotech applications 245 bipolar membrane 84, 107–109, 111 blending 32 block/graft copolymers 23 blood oxygenators 411 Bodenstein number 317 Boltzmann constant 7 boron content 458 boron problem 225 boundary conditions 316 bulk regime 69 bulk values 71 buoyancy 480

c cake-enhanced osmotic pressure 135 cake enhanced concentration polarization 130

Membrane Operations. Innovative Separations and Transformations. Edited by Enrico Drioli and Lidietta Giorno Copyright Ó 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32038-7

j Index

544

cake formation 125 calcination temperatures 51 capillary ﬁber membranes 170 capillary ﬁber modules 170 capillary ﬁbers 170 capillary membrane 170, 173 capillary membrane shapes 21 capillary modules 171 capsule membrane shapes 21 carbon-capture technologies 155 carbon-dioxide separation 179 Carman–Kozeny equation 322 cassette-type membrane contactors 501 casting solution 29 catalyst 215 catalyst immobilization 285 catalytic membranes XXIV, 285, 289, 301 catalytic membranes reactors 276, 285 catalytic reaction 278 catalytic shift reactors 188 cation-exchange 39 cation-exchange membrane 84 cell adhesion 435 cellulose acetate membranes 181 ceramic 209 ceramic capillaries 214 ceramic membrane materials for solvent separation 49 ceramic membranes 51, 52, 56, 206 ceramic oxygen-ion conductors 210 ceramic pervaporation membranes 49 cermet 209 cermet membranes 210 chain-immobilization factor 68 chain-packing procedures 7 chain ﬂexibility 22 charged ion-exchange membrane 26 chemical catalysts 285 chemical cleaning 378 chemical compatibility 215 chemical excess potential 9 chemical expansion 207 chemical expansion mismatch 215 chemical membrane reactors 287 chemical modiﬁcation 32 chemical reactions 296 chemical stability 24, 25, 210 chemical structure 22 chemometrics modeling 54 cleaning 378 clearance 416, 417, 420 closed-shell mode 324 co ions 84 CO2 capture 154, 195–197, 199, 206, 211, 212

CO2 greenhouse 231 CO2 permeance 211 CO2 removal 203, 205 CO2-selective membranes 204 CO2 selectivity 75, 157, 211 CO2 separation 211 CO2 separation membranes 211 CO2/H2 separations 188 CO2/N2 separations 186 coagulation bath 30 colloidal fouling 127, 131 commercial pervaporation membranes 48 commercially available MBR systems 386 COMPASS 8 composite membrane preparation 30 composite membranes 37, 63 composite skin layer 69 compression 202 computational ﬂuid dynamics 486 computer-aided molecular design (CAMD) 4 concentrates 229 concentration distribution 318 concentration modulus 323 concentration polarization 97, 98, 294 concentration-polarization layer 213 condensability 66 conditioning 65 conductivity 89 conﬁnement effects 70 constant of diffusion 5 continuous electrodeionization 113 continuous open-shell mode 324 controlled swelling 26 convection 55 convection-diffusion model 322 convective ﬂow proﬁle 323 convective velocity 318, 321 conventional activated sludge (CAS) 364 copolymerization 39 cost advantages 159 cost effective 191 cost of fouling 124 costs 155 counterﬂow 173, 174 counterﬂow module 175 counterﬂow/sweep membrane module 174 counterﬂow/sweep operation 173 counterions 84 creep 207 critical droplet diameter 484 critical ﬂux 125, 127, 376 cross ﬂow ﬁltration 428 crossﬂow membrane emulsiﬁcation 465, 476 crossﬂow module 174, 175

Index crosslinked polymers 207 cross-section structure 21 crystallinity 68 crystallization 450 current 111 current density 94, 99, 103 current utilization 100 cytocompatibility of membranes 436

d dehumidiﬁcation process 505 dense 207 dense ceramic 199 dense ceramic membranes 210 dense inorganic membranes 209 dense membranes 55 dense metal membranes 209 densiﬁcation 69 desalinated water quality 228 desalination 221, 230 desalination energy 231 desalination plant 226, 230 desalination process 229 dialysance 416 dialysate 428 dialysate diffusive resistance (RD) 420 dialysis–photocatalysis 351 dialyzer 415 differential mass-balance equation for statestate conditions 329 diffusion 55 diffusion based model 54 diffusion coefﬁcient 65, 66 diffusion constants 9 diffusion dialysis 115 diffusion selectivity 66, 68 diffusive transfer 416 diffusivity 207, 211 dip-coating 31 direct membrane emulsiﬁcation 465 disposable systems 261 distillation 450 Donnan dialysis 115 Donnan effect 352 downstream processing 245, 255 drag force 481 droplet-formation 479 droplet formation mechanism 485 droplets shape 486 dual-layer 160 dual-phase membrane 211 dual sorption model 65 dynamic lift 480 dynamic membrane emulsiﬁcation 466

e ecofriendly technologies XXIV electric current 89, 91 electrical potential 89 electrical resistance 85, 89, 104 electrochemical potential 91 electrodeionization 115 electrodialysis 83, 84, 94, 95, 102, 105, 107, 111, 222, 224, 250 electrodialysis equipment 100 electrodialysis reversal 101 electrodialysis stack 96 electrolysis 83 electrolyte solution 91 electromembrane 83 electromembrane process 89, 91 electron conductor membranes 210 electronics industry applications 490 ellipsometry 70 ellipsometry techniques 73 emerging membrane technology 216 emulsiﬁcation processes 468 emulsion liquid membranes 517 emulsions 463 end-stage renal disease 411 end-stage renal failure 413 energy 106, 139, 230 energy balance 296 energy consumption 105, 140, 187, 231 energy cost 142, 144, 145, 152, 230 energy dissipation 106 energy-efﬁciency 145 energy-efﬁcient 154 energy-efﬁcient membrane 144 energy-intensive 141 energy savings 160 energy source 224 energy systems 199 energy use 141 environment 229 environmental sustainability 139 enzymatic membrane reactors 404 enzyme immobilization 285, 405 enzyme membrane reactor 311 enzymes 285 equations 332 equivalent conductivity 90 ethanol dehydration technology 189 ethanol/water 191 evaporation-induced phase separation (EIPS) 27 external mass-transfer resistance 316 extracorporeal devices 427 extractive separations 516

j545

j Index

546

f fermentation 404 Ficks law 64 ﬁlm-forming properties 24 ﬁlm thickness 70 ﬁxed ionic charges 85 ﬁxed ionic moiety 85 ﬂat-sheet MBR 388 ﬂat-sheet membranes 63, 170 ﬂat sheet membrane shapes 21 Flory–Huggins equation 65 ﬂow-through applications 260 ﬂuid management 124 ﬂuorescence techniques 260 ﬂuorite 206 ﬂuorocarbon 39 ﬂux 53, 124, 206–210, 212, 225 food industry applications 488 force ﬁelds 5, 6 forces 480 forward osmosis 238 fouling 34, 121, 122, 126, 130, 235, 247, 374, 376 fouling causes 123 fouling mechanisms 125 fouling theory 125 fractionation 411 free volume 12, 13, 25, 69 functionalization 344 functionalized membrane 438

g gas diffusion 213 gas-permeability properties 74 gas-phase diffusion 212 gas-phase diffusion limitation 215 gas-separation XXIII, 167, 171, 208, 214 gas-separation membrane 63 gas-separation membrane modules 168 gases permeation 65 Gibbs free energy 108 glass-transition temperature (Tg) 23, 70 glassy gas-permeable membranes 183 glassy polymer membranes 64 glassy polymer speciﬁc volume (Vg) 65 glassy polymer(s) 23, 38, 64–66, 69, 71, 74, 175, 183, 207 glassy state 79 glassy state polymer 25 global balance models 479 global water market 385 good practice 226 gradient operation 6 green chemistry 335

greenhouse gas 195 greenhouse gas emission 216 Gusev–Suter transition-state method 9, 11, 12

h Ha number 317, 320 heavy hydrocarbons 182 hemodiaﬁltration 417, 418 hemodialysis 411, 413, 418 hemodialysis principles 414 hemodialyzer 411, 413, 414, 416, 418 hemoﬁltration 417, 418 Henrys law 64 hepatocyte cultures 435 heterogeneous catalysis 347 heterogeneous ion-exchange membranes 85, 86 heterogeneous photocatalysis 336, 346 high charge density 26 high-performance tangential ﬂow ﬁltration 256 high-pressure processes 130 high selectivity 79 hollow ﬁne ﬁber 169, 170, 214 hollow ﬁne ﬁber membranes 169 hollow ﬁne ﬁber membrane modules 180 hollow-ﬁber (HF) membrane bioreactor 439 hollow-ﬁber cartridge 418 hollow-ﬁber contactors 518, 525 hollow-ﬁber membrane shapes 21 homogeneous ion-exchange membranes 85, 86 homogenous anion 39 honeycomb elements 214 human hepatocyte bioreactor 439 human hepatocytes 436, 439 human lymphocyte membrane bioreactor 439 hybrid membranes 160 hybrid simulations 57 hybrid systems 354 hydrocarbon 39 hydrocarbon emission 186 hydrocarbon mixtures 191 hydrodynamic resistance 321 hydrogen-depleted layer 213 hydrogen diffusion 208 hydrogen ﬂux 209 hydrogen membrane reformer (HMR) 203, 215 hydrogen-permeable membranes 189 hydrogen permeability 210

Index hydrogen production 204, 345 hydrogen puriﬁcation 148 hydrogen-selective membranes 202, 213 hydrogen separation 175, 207, 208, 209 hydrogen-separation membranes 207 hydrophilic barriers 37 hydrophilic pervaporation membranes 47 hydrophilic polymers 37 hydrophilicity-hydrophobicity balance 24, 25 hydrophobic membranes 450 hydrothermal stability 208

i ideal membrane selectivity 143 immersed membrane modules 371 immobilization 405 industrial system 503 inertial force 480 inorganic CO2 selective membranes 211 inorganic membranes 208 inorganic nanoparticles 211 integrated membrane processes 265, 276 integrated membrane system (IS) 268, 274 integrated MF-NF-RO system 268 intensiﬁcation 404 interface models 7 interfacial polymerization 31 interfacial tension 480, 481 intermolecular forces 142 interpenetration 33 intrinsic ﬂuorescence probes 259 ion-exchange membrane 38, 39, 84, 85, 91 ions transport 88, 89 ion transport number 92

k Kedem–Katchalsky model 54 Knudsen diffusion coefﬁcient 453 Kozeny constant 322

l Laminating 31 L-aminoacids 403 large toxin removal 414 lattice Boltzmann 487 layer thickness 322 leather industry 273 Lennard–Jones potential 66 lifetime productivity 75 limiting current density 97, 99 limiting ﬂux 321 liquid-liquid extractors 451 liver-speciﬁc functions 437 L/L interfaces contactors 518

low electrical resistance 26 low-pressure process 126 lymphocytes 439

m macrosolute fouling 127 maintenance cleaning 378 mass balance 294 mass-balance equation 315 mass-transfer coefﬁcient 453 mass-transfer rate 318 mass transport 91, 246, 292, 325 mass-transport resistances 452 mass transfers 416 Maxwell–Stefan pervaporation 56 mean droplet diameter 479 mechanical properties 24, 85 mechanical stress 215 medical applications XXIII, 411 membrane aeration 381 membrane area 104 membrane-assisted integrated catalytic process 278 membrane-based solvent extraction (MBSE) 515, 519 membrane-based solvent stripping (MBSS) 519 membrane biocompatibility 414 membrane biohybrid system 436, 441 membrane bioreactor (MBR) XXIV, 224, 233, 234, 271, 285, 309, 310, 325, 363, 364, 366, 376, 385, 399, 402, 404, 438, 439 membrane-bioreactor basics 372 membrane bioreactor cleaning protocols 379 membrane-bioreactor conﬁgurations 368 membrane bioreactor hydraulics 373 membrane bioreactor plant(s) 371, 386 membrane chromatography 249, 260, 261 membrane cleaning 237, 378 membrane contactor(s) XXIV, 212, 350, 450, 452, 456, 499 membrane contactor applications 499 membrane contactor photoreactors 350 membrane counterion permselectivity 93 membrane crystallizers 459 membrane cutoff (l) 258 membrane distillation 238, 351, 450, 451 membrane distillation–photocatalysis 351 membrane emulsiﬁcation XXIV, 463, 464, 488 membrane emulsiﬁcation devices 476 membrane ﬁltration 381 membrane fouling 314 membrane gas-separation 63, 186, 191

j547

j Index

548

membrane integration 206 membrane mass-transfer coefﬁcient 453 membrane material 21, 251, 368 membrane microﬁltration 412 membrane modiﬁcation 32 membrane module 214 membrane permeability 203 membrane permselectivity 92 membrane photoreactor conﬁgurations 348 membrane photoreactors 348 membrane plasmapheresis 421 membrane preparation 26 membrane processes XXIII membrane properties 207 membrane reactors (MRs) XXIV, 199, 202, 285, 287, 294 membrane-replacement 383 membrane resistance 125, 141 membrane(s) 19, 48, 206, 215, 222, 224, 289, 310, 311, 352, 411, 419, 434–436, 468 membrane selectivity 205 membranes energy 139 membrane-separation platform 158 membrane separation units 157 membrane sewage treatment (MST) 365 membrane shapes 21 membranes impacts 139 membranes issues 139 membrane solvent pervaporation 45 membrane solvent separation 45 membrane systems 195 membrane technology 365 membrane types 168 metal 209 metal membranes 209 methane 179 methane steam reforming (MSR) 202, 215 methane steam reforming reaction 215 microcapsules 474 microﬁltration (MF) 141, 327, 350 microﬁltration membranes 223, 224, 365, 421 microporous carbon 208 microporous membrane 208, 428 microporous polymeric ﬂat-sheet membrane 500 microporous silica membranes 208 mixed-matrix membrane(s) 32, 37, 63 mixing residue gas 175 mobility selectivity 143, 144 modeling 314, 327 modelling XXIII modiﬁed membranes 437 module 170

module conﬁgurations 168 molar conductivity 90 molar mass 22 molecular dynamic 6 molecular modeling 3, 5 molecular modeling tools 4 monolithic membranes 215 Monte Carlo simulation method 10 moving-bed bioﬁlm membrane reactor 312 multichannel elements 214 multieffect distillation 238

n nanoﬁltration (NF) XXIII, 36, 46, 52, 253, 261 nanoﬁltration-coupled catalysis 279 nanoﬁltration ﬂux models 53 nanoﬁltration membranes 223, 224 nanoﬁltration rejection 55 natural-gas treatment 178 Navier–Stokes approximation 332 Nernst–Planck ﬂux equation 91 neuronal network 441 neuronal tissue-engineered constructs 440 neuronal tissue reconstruction membranes 440 Newtons equations of motion 6 nitrogen air 171 nonsolvent-induced phase separation (NIPS) 27 novel photocatalytic membranes 279

o O2/N2 selectivities 171 Ohms law 89 oleﬁn/parafﬁn separation 150 oleﬁn-parafﬁn separation membranes 149 operating conditions 452 operating costs 105, 111 operating mode 126 optical resolution 402 optimal packing density 420 optimizing membrane bioreactor 383 organic solvents 45, 48 organic vapor mixtures separation 191 organophilic barriers 37 organophilic pervaporation 251, 253 organophilic pervaporation membranes 38 organoselective barriers 37 organoselective membranes 38 osmotic distillation 275 overall clearance 418 oxy-fuel 196, 206 oxy-fuel process 197 oxygen separation 206

Index oxygen-separation membranes 207, 210 oxygen transfer rate 380

p packing density 206 parallel-plate hemodialyzers 419 partial oxidation 199 partial oxidation reactions 290 partial photocatalytic oxidations 354 particle-deletion algorithm, DPD 9 particulate fouling 126, 130 partition coefﬁcient 143 partitioning 142 partitioning selectivity 143, 144 Pd-based composite membranes 209 Pd-based membranes 204, 209, 210, 213, 290 Pd metal 204 Peclet (Pe) number 317, 319 penetrant size 66, 79 penetrant solubility 79 perﬂuorocarbon polymers 87 peritoneal dialysis 413 permeability 5, 13, 14, 55, 68, 69, 74, 172, 211, 292 permeability coefﬁcient (P) 64, 65 permeability-thickness relation 204 permeable membranes 214 permeance 188, 208, 209, 214, 290, 294 permeation 73, 294 permeation driving force 299 permselectivity 66, 76, 77, 85 perovskite 206, 207 pertraction 516, 528 pervaporation 37, 46, 52, 56, 189, 251, 278 pervaporation membranes 48 pervaporation–photocatalysis 351 pH of the aqueous solutions 340 pharmaceutical industry applications 489 phase inversion 27 phase separation 27 photocatalysis 335 photocatalysis applications 341 photocatalysts 336, 338 photocatalytic 335 photocatalytic membrane reactors 348, 352, 353 photocatalytic oxidation of benzene 357 photocatalytic parameters 340 photocatalytic processes 341, 347 photocatalytic technologies 346 photodegradation 353 photodegradation efﬁciency 349

photografting 33 photoreactors 347 physical aging 64, 69, 71, 75 physical cleaning 378 plasma collection 412 plasma exchange 423 plasma polymerization 31 plasma separation 411, 412, 421, 422 plasticization 65, 68, 122, 191 polluted water 232 polymer-based membranes 207 polymeric membrane 19, 21, 22, 35, 48, 52, 53, 191, 211 polymerization reactor 184 polymer membranes 34 polymer packing models 6 polymers 22, 23, 25–27 polymers intrinsic microporosity (PIMs) 38 pore plugging 125 pores closure 122, 125 porous 207 porous membrane 212 positron annihilation lifetime spectroscopy (PALS) methodology 13 postcombustion 211 postcombustion capture 196, 205, 211 postcombustion processes 196 power generation 154 precombustion 203, 211 precombustion CO2 capture 199 precombustion decarbonization 196, 202, 206, 207 precombustion processes 204 premix membrane emulsiﬁcation 465 pressure exchangers 230 pressurized membrane photoreactors 349 pretreatment 123, 128, 236 prevention 236 process conﬁgurations 371 process design 113 process intensiﬁcation XXIV, 265, 287, 456 product quality 225, 234 production process 404 productivity 63, 73 propylene/propane separation 152 protein mass ratio 258 proteins 255 proto-membrane 27, 30 proton exchange membrane 208

q quadrapole interaction 75 quiescent condition 465, 485

j549

j Index

550

r reactor design 215 recovery level 236 red water 228 redox potentials 337 reduction reactions 344 reﬂection coefﬁcient 55 reﬂection curve 55 refractive index 70 refractive index change 73 refrigerated storage cells 507 regenerative medicine 433 rejection 55, 352 relaxation 383 removal of volatiles/gases 451 repulsive interactions 352 required membrane area 111 resistance 103, 215 resistance-in-series model 53, 322 respiratory quotient (RQ) 78 reverse osmosis 36, 222, 365, 385 reverse osmosis biofouling 132 reverse osmosis desalination 224, 266 reverse-osmosis desalination costs 224, 226 reverse-osmosis desalination process 226 reverse-osmosis membranes 222 reverse osmosis particulate fouling 130 reversible electrodialysis 224 Robeson plot 172, 181, 191 rotating membrane emulsiﬁcation 476 rubbery membranes 182 rubbery polymers 13, 23, 25, 64–66, 75, 79, 182, 183, 207 rubbery polymers size selectivity 75 rubbery polymers vapor sorption isotherm 65 rubbery vapor-permeable membranes 184

s salt rejection 225 sanitation 385 scale formation 130, 133 selective barrier structure 19 selective layer 74, 212 selective membranes 187 selective oxidations 343 selective plasma puriﬁcation by cascade ﬁltration 423 selective therapeutic plasmapheresis 428 selectivity 63, 68, 73, 79, 172, 188, 208, 211, 249, 292 self-assembly 36 semiconductor 336 semicrystalline polymer 23, 68 semipermeable membrane 428, 434

separation 45, 206, 208, 209, 210 separation energy efﬁciency 139 separation factor 141 separation mechanism 207 sewage water 232 side-stream conﬁguration 371 side-stream membrane modules 371 sieving mechanism 19 silica-based zeolite 208 silica membranes 208 size exclusion 208 skins 69 small bioactive molecules 249, 250 solar energy 353 solubility 5, 9, 207, 211 solubility coefﬁcient 9, 66 solubility selectivity 66, 75, 76 solute transport 54, 55 solution-diffusion mechanism 53, 64, 66 solution-diffusion model 54, 55 solution-diffusion pervaporation 56 solvation intermolecular forces 143 solvent/nonsolvent system 29 solvent-resistant nanoﬁltration 47 solvent stability 52 solvent transport 54 sorption-diffusion-desorption process 56 sorption-diffusion models 56 sorption of gases 65 speciﬁc aeration demand 381 speciﬁc conductivity 89 speciﬁc desalination energy 105 speciﬁc volume 64 specimen size 69 Spiegler–Kedem model 54 spiral-wound modules 170 spiral wound membrane modules 180 stability 207, 208, 209, 210 stack design 108 static membrane emulsiﬁcation 466 stationary permeate ﬂux 322 steam reforming 202 stimulus-responsive membranes 248 stirred membrane emulsiﬁcation 465 stirred-tank reactor 327 stirred vessel 476 stripper/scrubber 451 submerged membrane photocatalysis reactor 350 submerged membrane photoreactors 349 suction of permeate 324 sulfur resistance 210 supported liquid membranes 450 surface kinetics 212

Index surface layer/cake 123 surface modiﬁcation 32, 33 suspensions 463 sustainable ﬂux 125, 377 sustainable growth 265 sustainable technical solutions 265 swelling 56, 85 synthetic membranes 19

transport number 92 transport simulation/prediction 8 tubular membrane shapes 21 tubular/hollow-ﬁber membrane bioreactor 388 tunnel mechanism 93 turbochargers 230

t

ultraﬁltration 34, 141, 223, 256, 261, 327, 350 ultraﬁltration membranes 365 uniform transmembrane pressure 247 unmodiﬁed membranes 437 unused energy 144

therapeutic 422 therapeutic plasmapheresis 423 thermal expansion 207 thermal expansion coefﬁcient 209 thermal expansion mismatch 215 thermally driven processes 149 thermally induced phase separation (TIPS) 27 thermal stability 24 thermodynamic inefﬁciency 145 thermodynamic potential gradients 207 thickness 70, 212 Thiele modulus 317 thin dense layer 64 thin dense skin 64 thin-ﬁlm composite (TFC) 21 thin-ﬁlm composite membranes 36 thin ﬁlms 69, 70 thin Pd layers 209 thin selective membrane layers 212 tissue engineering 433 titanium dioxide 338 torque balance 483 tortuosity 68 total current 103 total energy 105, 111 total energy consumption 112 total membrane area 104 total oxidations 341 track-etching 26 transformation 285 transition temperature (Tg) 64 transmittance 418 transport 19, 213 transport mechanism 56 transport models 54

u

v vapor/gas separations 183 vapor-induced phase separation (VIPS) 27 voltage drop 104 volume fraction 68 volume relaxation 69

w wafer-like 215 waste water treatment 365 waste water treatment by whole-cell membrane reactor 313 waste water-treatment process 233 waste water treatment membrane 364 water treatment membrane techniques 221 water desalination XXIII, 221, 456 water dissociation equilibrium 107 water gas shift (WGS) 202, 215 water gas shift reactor 154 water quality 222 wavelength/light intensity 341 whole-cell membrane bioreactor 312 Widom method 9 work exchangers 230

y yearly growth 385 Young–Laplace 480

z zeolite 208

j551